March 14, 2012

Industry Report

Technology: Enterprise IT Infrastructure

Storage 3.0: Storage Innovation Enters An Exciting New Phase
THINK SUMMARY: We believe an exciting new wave of enterprise storage innovation is upon us driven by a new generation of interesting startups that are leveraging emerging new technologies to solve growing problems, address new workloads and, in the process, bust several myths that have defined the storage industry until today. We attempt to examine several trends in this paper that we believe could potentially impact the industry significantly in the near and medium term and also profile the private companies participating in this innovation. KEY POINTS: We believe innovation in processors (multi-core), compute buses (PCI 3.0), networks (10Gb Ethernet), and memory (NAND Flash) have outpaced proprietary storage controllers over the last decade and created opportunity for companies willing and untethered by legacy architectures and deployments. Almost simultaneously the growing data load in the enterprise, greater demands on power efficiency in the data center and new emerging workloads such as VDI and Big Data, are creating the need for a major rethink in storage. Innovation is resulting in several assumptions being shown the door and spawning new product-market sub-segments within storage. We look at some of these long-standing assumptions being shown the door in this paper and emerging trends driving this change. Concepts being spawned or making a comeback include DAS, converged infrastructure, converged primary and backup storage and performance provisioning. This paper features interviews with leaders of companies participating in this innovation. We also look at the respective value propositions of these players, their strategies and workloads they seek to address in this paper. Like always, we expect competition from legacy Storage vendors to heat up in the near-term. EMC with the announcement of VFCache and Project Thunder has signaled its intent. Hardly a day goes by before we hear a new product announcement from an existing large vendor or a private company. Ideas are coming out thick and fast, but we believe the real relative winners in this exciting dynamic are likely to be players that execute relentlessly and effectively not only in product development but also in go-to-market.

Reason for Report: Industry Update

Rajesh Ghai, CFA 415-249-6365, [email protected] Ryan Brookman 415-249-1978, [email protected]

Please see analyst certification (Reg. AC) and other important disclosures on pages 170-171 of this report.

March 14, 2012

Industry Report

Data Center Infrastructure

Storage 3.0: Storage Innovation Enters An Exciting New Phase

Source: TechCrunch

Rajesh Ghai, CFA 415.249.6365 [email protected] Ryan Brookman 415.249.1978 [email protected]

Please see analyst certification (Reg. AC) and other important disclosures on pages 170-171 of this report.

Page 2

March 14, 2012

Industry Report

Table of Contents 1.0 Executive Summary ........................................................................................................................................................... 4 2.0 The Storage Landscape Today ......................................................................................................................................... 5 3.0 The Technology Enablers .................................................................................................................................................. 8 3.1 NAND Flash ......................................................................................................................................................... 8 3.2 Multi-Core Processors .......................................................................................................................................14 3.3 10Gb Ethernet ....................................................................................................................................................16 3.4 Server, Storage and Desktop Virtualization .......................................................................................................17 3.5 Virtual Desktop Infrastructure (VDI) ...................................................................................................................18 3.6 The Cloud ..........................................................................................................................................................21 4.0 Storage Myths Being Busted! ..........................................................................................................................................24 5.0 New Storage Product Markets Being Spawned ..............................................................................................................27 Product Market Snapshot ..............................................................................................................................................28 5.1 Server-Side Flash ..............................................................................................................................................29 5.2 All-Flash Storage Arrays ....................................................................................................................................33 5.3 Flash-Optimized Hybrid Storage Arrays ............................................................................................................38 5.4 Converged Infrastructure (Leveraging Flash Storage) ......................................................................................41 5.5 Ethernet SAN .....................................................................................................................................................42 5.6 Flash-Accelerated Storage Optimization Appliances ........................................................................................43 6.0 Conclusions .....................................................................................................................................................................44 7.0 Interviews with Select Private Company Leaders ...........................................................................................................45 Ron Bianchini, CEO and Co-Founder of Avere Systems .............................................................................................46 Kevin Brown, CEO of Coraid .........................................................................................................................................50 Jean-Luc Chatelain, Executive Vice President, Strategy and Technology of DataDirect Networks .............................54 Gareth Taube, VP of Marketing at Kaminario Technologies ........................................................................................57 John Spiers, CEO and Founder of NexGen Storage ....................................................................................................60 Suresh Vasudevan, CEO of Nimble Storage ................................................................................................................64 Tom Isakovich, CEO and Founder of Nimbus Data Systems .......................................................................................70 Scott Genereux, President and CEO of Nirvanix ..........................................................................................................75 Dheeraj Pandey, Founder and CEO of Nutanix ............................................................................................................80 Scott Dietzen, CEO of Pure Storage .............................................................................................................................83 Dave Wright, CEO and Founder of SolidFire ................................................................................................................87 Ursheet Parikh, CEO and Founder of StorSimple ........................................................................................................91 Dan Scheel, President of Texas Memory Systems ......................................................................................................97 Kieran Harty, CEO and Co-Founder of Tintri ..............................................................................................................101 Daniel Crain, CEO of WhipTail....................................................................................................................................105 Appendices ..........................................................................................................................................................................108 Appendix I: Private Company Profiles ........................................................................................................................109 Appendix II: Venture Capital Profiles ..........................................................................................................................149 Appendix III: Glossary of Technical Terms .................................................................................................................167

Page 3

March 14, 2012

Industry Report

1.0

Executive Summary

We believe an exciting new wave of enterprise storage innovation is upon us driven by a new generation of promising startups that are leveraging emerging new technologies to solve growing problems, address new workloads and in the process bust several myths that have defined the storage industry until today. We attempt to examine several trends in this paper that we believe could potentially impact the industry significantly in the near and medium term and also profile the private companies participating in this innovation. We believe innovation in processors (multi-core), compute buses (PCI 3.0), networks (10Gb Ethernet), and memory (NAND Flash) have outpaced proprietary storage controllers over the last decade and created opportunity for companies willing and untethered by legacy architectures and deployments. Almost simultaneously the growing data load in the enterprise, greater demands on power efficiency in the data center and new emerging workloads such as VDI and Big Data, are creating the need for a major rethink in Storage. Innovation is resulting in several assumptions being shown the door and spawning new product-market sub-segments within storage. We look at some of these long-standing assumptions being shown the door in this paper and emerging trends that we believe are driving this change. Concepts being spawned or making a comeback include DAS, converged infrastructure, converged primary and backup storage and performance provisioning. The following represents our attempt to categorize the new companies profiled in this paper along some emerging new product-market sub-segments: Server-side Flash Storage: Fusion-io, Virident, STEC, OCZ All-Flash Storage Arrays o High-performance market focus: Kaminario, Violin Memory o Enterprise market focus: Nimbus, Pure Storage, WhipTail o Cloud Service provider market focus: SolidFire Hybrid Flash-optimized Storage Arrays: DDN, NexGen, Nimble Storage, Tintri Ethernet SAN: Coraid Converged Infrastructure: Nutanix Storage Optimization Appliances o Performance Optimization: Avere Systems, CacheIQ, Dataram, GridIron o Cloud Gateway: StorSimple Cloud Storage: Nirvanix

This paper features interviews with leaders of companies in Bold in the list above. We also look at the respective value propositions of these players, their strategies and workloads they seek to address in this paper. Like always, we expect competition from legacy Storage vendors to heat up in the near-term. EMC with the announcement of VFCache and Project Thunder has signaled its intent to take on the first two product-market segments listed above. Hardly a day goes by before we hear a new product announcement from an existing large vendor or a private company. Ideas are coming out thick and fast, but we believe the real relative winners in this exciting dynamic are likely to be players that execute relentlessly and effectively not only in product development but also in go-to-market. That said, we look forward to tracking this space closely and identifying the winners of tomorrows Storage market.

Page 4

March 14, 2012

Industry Report

2.0

The Storage Landscape Today Data Growth

We expect data growth to accelerate as the global economy continues its recovery in 2012. We believe this acceleration is likely to come from three main sources. First, the digitization of new business processes especially in the developing world (BRIC countries), which is likely to drive demand for new IT projects and hence resources to store and carry the new data. The second source of data traffic growth, in our opinion, is likely to come from the rapid proliferation of mobile devices and, hence, user developed content on the Internet, such as video and photos and images. The third source is likely to be the incremental increase in data produced as the velocity of business accelerates, in our view. According to the findings of the IDC study, The Digital Universe Decade Are You Ready? published in May 2011, the total amount of stored data is likely to grow 44 times between 2009 and 2020 (see Exhibit 1). This implies that stored data is likely to grow at a CAGR of 40% over the next 10 years. In todays context, it is also surprising to us that IDC expects digital information to grow 50% Y/Y in 2010 from 0.8 Zb in 2009 to 1.2 Zb in 2010. A zettabyte (Zb) is a million petabytes, which, in turn, is a trillion gigabytes. More than 70% of the digital universe is generated by individuals. But enterprises have responsibility for the storage, protection, and management of 80% of the digital universe. This enterprise liability will only increase as social networking and Web 2.0 technologies continue to invade the enterprise. Exhibit 1: Growth of Data Storage, 2009- 2020E

Source: IDC Increasing importance of data center power efficiency We believe as the amount of data continues to grow in the enterprise, it is inevitable that it will drive the need for even greater compute, network and storage capacity in the enterprise. Indeed, a number of studies conducted by various industry groups suggest that this will necessitate that enterprises spend an order of magnitude higher number just to keep this infrastructure running in terms of increased power, cooling and floor space. To cite one such study conducted by the EPA (Exhibit 3), despite the deep recession in 2009 and 2010, data center energy consumption in the US was expected to increase by almost 50% between 2008 and 2011. According to the EPA, Data centers

Page 5

March 14, 2012

Industry Report

consume 1.5% of all U.S electricity for a total annualized cost of $4.5 billion. By 2011 the annual electricity cost for data centers was estimated to reach $7.4 billion. Over 40% of the power used by data centers goes to cooling costs. Several studies indicate that in the coming years energy costs will consume up to one-third of IT budgets. Exhibit 2: US Data Center Usage Projections

Source: EPA We believe a consolidation of data centers and within data centers a consolidation of infrastructure enabled by virtualization is a likely panacea to this problem. However, even with a move to a cloud computing paradigm, we believe enterprises will need to spend a lot more on energy, as has been suggested in projections by Pike Research (Exhibit 3). Exhibit 3: US Data Center Usage Projections by the EPA

Source: Pike Research

Page 6

March 14, 2012

Industry Report

New Workloads Several new storage workloads have come up specifically in the past few years which have placed new demands and created needs that were unmet by traditional enterprise storage systems. Some workloads that we believe represent new challenges for storage include: o Virtual Desktop Infrastructure (VDI) This workload is characterized by a high incidence of random block I/O which has the potential to exacerbate the latency of spinning disk while serving Enterprise scale users Web-scale Clustered Environments These environments typically run single application homogenous environments for their millions of concurrent users on thousands of clustered commodity servers. Considering the scale of operations in these environments and the amount of hardware capacity that is being added very frequently, web-scale environments are characterized by distributed scale-out file systems that typically leverage Direct Attached Object based Storage. Big Data Analytics/Hadoop Clustered distributed file system based implementations that leverage commodity hardware to provide the requisite scale and performance necessary to analyze petabyte scale data stores.

Page 7

March 14, 2012

Industry Report

3.0

The Technology Enablers

In this section we look at a few specific technologies that are enabling the paradigm shift that is currently on in storage today. 3.1 NAND Flash

NAND Flash, in our view, has the potential to revolutionize Enterprise Storage and we believe this is an understatement. We believe the magnitude of change and its impact could be similar or greater than the impact when disk based storage th began to replace tape in the late 20 century. Long associated with consumer electronics, NAND Flash is fast becoming a viable storage medium for enterprise applications in the data center. It is well established by now that Flash based solid-state storage solutions outperform storage systems based on hard disk drives (HDDs), even when HDDs are combined into large, multi-disk, striped arrays, short-stroked, or managed in other ways to increase performance. Less well known, however, due to some lingering misperceptions about NAND Flash as a storage medium, is the fact that NAND Flash-based storage doesnt just outperform its mechanical counterpartit also significantly improves on data integrity (confidence that what you put in for storage is exactly what you get out) and availability (confidence that your data can be retrieved whenever you want it), when the appropriate firmware and Flash Controllers are utilized to read and write from Flash. Writing data to a Flash chip involves three main operations: read, erase, and program. Reads can occur randomly anywhere within the 16,384 blocks of a typical Flash chip. Erasing sets all of the bits in an individual Flash block to 1. Programming only occurs on blocks that are erased, because the program operation can only change a bit from a 1 to a 0. The high voltages needed for the erase operation originally gave Flash its name. Flash drives targeted toward the consumer market in thumb drives, cameras, media players, etc., accept the limited endurance or write cycles of Flash and its poor write performance because writes occur fairly infrequently and therefore dont pose a problem. In the enterprise, however, the erase stress and endurance, coupled with slow write performance and some data integrity issues, explain why until recently Flash drives were seldom deployed in mission critical enterprise environments. As a storage medium, NAND Flash offers some unique challenges, including bit errors, wear-out, and part failures, particularly during infant mortality. The voltage required to set the value of a NAND Flash cell, combined with the everdecreasing size of the cells, creates some probability that there will be errors, and that some cells will become unreliable with time and use. Errors can occur for a large number of reasons many of them random occurrences, but the small (and ever decreasing) size of NAND Flash gates makes them particularly susceptible to error and failure. One such example is read or program disturbs. This is when reading or programming the data in one cell inadvertently disturbs the data in a nearby cell. NAND Flash errors can also be caused by elevated heat, manufacturing defects, or even simply repeated use, also known as wear out. Another peculiarity of NAND Flash as a storage medium is its increasing susceptibility to bit errors after having gone through a certain number of program-erase (P/E) cycles. The number of cycles varies, depending on density (storage capacity per physical area), vendor, and NAND Flash type. Generally speaking, SLC is usually rated for ~100,000 P/E cycles, whereas MLC Flash is usually rated for ~5,000-10,000 P/E cycles. NAND Flash also becomes less reliable over time when unpowered. While NAND Flash allows very fast read times, programming and erasing data take comparatively longer. Programs take ten times longer than reads (hundreds of microseconds) and erases take roughly 100 times longer than reads (multiple milliseconds). Therefore, the current state of Solid State Storage technology must deal with asymmetrical access. Innovation in enterprise flash storage has focused on developing controller strategies to deal with these limitations and issues and to improve performance. While NAND Flash may present some unique challenges, both the hardware (controllers) and the software (firmware, management utilities) that manages the read, erase and the write processes, when designed properly, can compensate for those challenges. Two specific technologies that go a long way to enhance the endurance, reliability and life of Flash and make it ready for the Enterprise are Error Correction Codes (ECC) and Wear Leveling. ECC can greatly improve data integrity even when there are bit errors. Wear leveling essentially spreads data evenly across all flash chips to spread the damage caused by erase and write cycles across the available flash. A good controller can more than compensate for

Page 8

March 14, 2012

Industry Report

the peculiarities of Flash while bringing out its primary strengths compared to HDDs: performance (bandwidth speed, lower latency), lack of mechanical parts which results in lower maintenance and repair costs, and comparatively low energy use. NAND Flash Performance Advantage As is evident from the chart below HDD performance in terms of latency and bandwidth improvements, has plateaued over the past decade in particular. The impact of this HDD performance stagnation has been exacerbated by the fact that CPU speeds have in accordance with Moores Law continued to see an exponential increase in data processing speeds over the same period. We believe the mismatch between data processing speeds of CPUs and data supply speeds of HDDs has created a major data supply problem that is getting worse as CPU vendors INTC and AMD continue to release new faster platforms every 18 months. This data supply problem in our view screams for the need for Silicon based storage that can potentially bridge the gap with CPU speeds and alleviate this growing data supply problem in the data center. We believe NAND Flash based storage has the potential to step into this void quite successfully. We believe datacenter CPU utilization rates are constrained by this data supply problem more than anything and the use of Flash can potentially consolidate more servers and reduce the overall TCO (total cost of ownership) of the datacenter.

Exhibit 4: HDD Latency and Bandwidth Improvements Have Plateaued

Source: IBM Research

The NAND Flash performance advantage becomes even more evident when it comes to providing storage to mixed read/write applications such as e-mail, VDI or large database applications, which require random I/O. Latency in applications which require random I/O becomes particularly pronounced when writing random blocks to disk, with each write suffering the average latency associated with HDDs. Enterprise-grade SSDs are particularly well-suited to handle an extremely high volume of small block random I/O transfer operations and are a good fit for these applications. Todays enterprise and server-class data center applications that require more than 100,000 IOPS, such as web-scale systems such as those at the large Internet properties and multi-million record databases, are becoming very common. PCIe based solid state storage solutions such as those from Fusion-io have found early success in satisfying the operational requirements of these applications. NAND Flash Reliability Advantage HDDs are the number one failing component being replaced in storage systems worldwide, causing an overall increase in TCO. This is mainly a result of the fact that HDDs are spinning devices and are hence prone to mechanical failure caused by vibration. According to studies conducted at CMU, the overall failure rates in controlled environments show ranges

Page 9

March 14, 2012

Industry Report

from 2% to 8.6%, based on existing studies and industry data. An annual failure rate of 2 to 8% indicates that as many as one out of every 12 HDDs deployed will fail every year. Replacement cost can be much higher than just the cost of the drive itself, when the costs of service personnel and system downtime are taken into consideration. Dependent on manufacturer and design, solid state storage devices can outperform HDDs in all of these areas, saving time, money and resources required for storage system maintenance. NAND Flash Power and Cooling Advantage According to the EPA, Data centers consume 1.5% of all U.S electricity for a total annualized cost of $4.5 billion, by 2011 the annual electricity cost for data centers was estimated to reach $7.4 billion, and over 40% of the power used by data centers goes to cooling costs. Several EPA studies indicate that in the coming years, energy costs will consume up to one-third of IT budgets. Because solid state devices have no rotating media, read/write heads, actuators, or spinning motors, it is easy to understand why power consumption in SSDs is substantially lower than those for HDDs. In some Tier 0 & 1 storage systems, SSDs can save over 80% in total storage system energy requirements. NAND Flash Cost/Gb progression The only offsetting factor that has prevented wider adoption of NAND Flash-based solid state storage has been relative acquisition cost/unit of storage compared to HDDs. Enterprise-grade SSDs costs roughly 3-4x more than enterprise-class HDD in terms of cost/Gb. When comparing the acquisition cost of SSD vs. HDDs in a high random I/O (input/output) transaction application (i.e. email, other transaction based applications.), an SSD device can be used to replace an array of 10 or more HDDs, providing a much smaller footprint, higher performance, and lower hardware and software licensing costs. In the past, the main barrier to the deployment of SSDs has been their high cost per gigabyte. But the cost barrier is diminishing rapidly, with flash prices showing an average 30-40% yearly decline over the past decade. Flash manufacturers continue to push product roadmaps to increase density and reduce costs. For example, Samsung has reduced its process geometry from 63nm in 2006 to 32nm in 2009 to 2xnm in 2011, resulting in increasing densities per chip and reduced cost per gigabyte. We believe this dynamic will result in NAND Flash coming down the cost curve more steeply in coming years as the chart below indicates. While we do not believe HDDs will get completely replaced by SSDs in enterprise storage it is never a zero-sum game, tape still exists we do believe NAND Flash based storage will begin to significantly replace the performance tier of HDDs in coming years. These drives include the Fibre Channel and SAS interface drives. Our research suggests these HDD today sell for $2-3/Gb which is much closer to the $4-5/Gb price point that MLC-based drives from vendors such as STEC are likely to make available later this year, in our view. We do expect HDDs to continue to dominate the capacity tier of storage with the SATA interface drives. Storage system vendors have typically adopted tiered storage approaches in their systems while trying to leverage the performance of SSDs and performance HDDs for performance and SATA HDDs for capacity. We believe over time SSDs should replace performance HDDs altogether and lead to only two kinds of storage drives in Enterprise storage systems to quote Larry Ellison of Oracle Flash and Trash (SATA HDDs)!

Page 10

March 14, 2012

Industry Report

Exhibit 5: Cost/Gb Progression of Storage Media

Source: SNIA, IBM Research

Solid State Storage Form Factors Enterprise Storage Systems utilize NAND Flash in one of three form factors - Drive, Card or Module. 1. Solid State Drive (SSD) A SSD can be used to replace a HDD without too much physical change in the adjacent physical environment and hence comes in traditional HDD form factors such as 5.25-inch, 3.5-inch, 2.5-inch or 1.8-inch and traditional storage protocol interfaces such as Fibre Channel, SAS and SATA. The largest size, 5.25-inch, is not widely used, except for special-purpose appliances, such as backup devices. 1.8inch SSDs are used in removable and ultra-mobile applications. 3.5-inch drives are mainly used in desktop PCs and enterprise systems. Portable applications, such as laptops utilize 2.5-inch drives. 3.5inch and 2.5-inch are the most common HDD form factors and the same holds true for SSDs, as they will fit into the same slots as the identical size HDDs. SSDs are defined by the HDD storage protocol interface that is used to connect to the drive such as Fibre Channel, SAS or SATA and usually utilize a storage protocol conversion controller. Vendors supplying SSDs to enterprise storage systems include STEC, Samsung and Hitachi GST.

Page 11

March 14, 2012

Industry Report

Exhibit 6: SSD

Source: STEC, Inc. 2. Solid State Card (SSC) A SSC resides on a printed circuit board and can utilize a standard card form factor such as a PCIe card. A solid state card would typically use an interface such as PCIe. Compared to a solid state module, an SSC normally has a larger physical size, more capacity, and higher performance. Vendors selling systems based on PCIe cards today include Fusion-io. Exhibit 7: Solid State Card

Source: Fusion-io

Page 12

March 14, 2012

Industry Report

3. Solid State Module (SSM) A SSM resides in a dual in-line memory module (DIMM) or similar form factor and may use a standard HDD interface such as SATA and in some cases a PCIe standard. A solid state module is typically smaller in physical size and has lower capacity and performance than a solid state card. It is typically used in a storage array with a controller designed to manage NAND Flash based DIMM, or VIMM (Virtual Intelligent Memory Module). Violin Memory utilizes a SSM called VIMM in its all-Flash storage arrays. Exhibit 8: Solid State Module

Source: Violin Memory

Page 13

March 14, 2012

Industry Report

3.2

Multi-Core Processors

A key factor enabling consolidation of compute resources on a smaller hardware footprint has been the advent of multicore server processors in 2009. Intel launched its quad-core server CPU Nehalem in June 2009 and AMD followed soon after with its Opteron family. 2010 saw the advent of even higher-core processors from Intel (Xeon 7500), AMD and IBM (POWER7). Multi-core processors, starting with the dual-core processor, are the answer to Moores Law running into the brick wall of physical reality. Specifically, traditional approaches of increasing processor performance by moving to higher clock rates eventually ran into power consumption and expensive cooling issues. Multi-processor systems provided an interim solution. Multi-core processors have the advantages of a smaller footprint with greater performance per watt. This makes multi-core processors ideal for traditional server architectures and particularly well suited for blades and other highdensity computing needs where space and cooling are at a premium. Besides providing much higher compute capacity than a single core processor could, we believe the advent of multi-core processors has also provided a big boost to deeper penetration of server virtualization in the data center. For example, a quad-core Nehalem processor based server could now host four virtual machines on the same physical server one on each core - thus consolidating four workloads that were running on four different physical servers on one physical server. Server consolidation has resulted in a much higher compute density in the data center. The key implication of this development from the point is that each physical server is clearly processing more data and generating orders of magnitude higher network traffic as a consequence. This exacerbates the high-latency and stagnating speed problems that are synonymous with HDDs today and have created the need for new storage paradigms such as Flash-based storage.

Exhibit 9: Server Consolidation Using Virtualization on Multi-Core Processor-Based Servers

Source: d3planet.com Servers based on multi-core processors also allow data centers to grow compute power without growing space or cooling requirements. Replacing older single-core servers with multi-core servers can deliver three to five times the compute power within the same hardware footprint. Even greater efficiencies can be achieved by consolidating applications onto fewer, more powerful servers using virtualization. The multiplied capabilities and efficiency of multi-core processor based systems also raises the demand for I/O capacity. While the multi-core server does provide ample headroom for

Page 14

March 14, 2012

Industry Report

consolidating multiple applications onto the server, the aggregation of application I/O traffic can easily require the additional bandwidth of 10GbE connectivity for optimum network performance. Besides the impact on server systems in the datacenter, we believe the enhanced horsepower of multi-core processors provide new processing capability to storage array controllers that will allow even more function and intelligence to be designed in them. We believe new software capabilities such as in-line de-duplication, predictive failure analysis, automated tiering, and intelligent drive spin-down have the potential to take array controllers to new heights in the future. In addition, we believe with multi-core processors it will be possible to scale storage systems further by creating the ability to manage several petabytes of storage via a smaller controller footprint. In addition, we believe, the multi-threading possible with multi-core processors is also likely to enable a greater diversity of workloads to be supported simultaneously within a single controller, and allow support for variable block length data streams thus improving the flexibility and versatility of storage systems immensely. We already see some of the new innovative storage players profiled in this whitepaper such as Nimble and Tintri exploiting this technology advancement. Exhibit 10: Support for Variable Block-Length Storage Using Intel Multi-Core Processors

Source: Nimble Storage

Page 15

March 14, 2012

Industry Report

3.3

10Gb Ethernet

Since ratification of the first 10 Gigabit Ethernet (10GbE) standard in 2002, 10GbE usage has grown primarily in niche markets demanding the highest bandwidth available. This established 10GbE as a stable, standards-based connectivity technology and the next evolutionary stage above Gigabit Ethernet (GbE). However, connectivity needs today extend beyond just providing a higher bandwidth pipe. Certainly, bandwidth and I/O throughput are important aspects of datacenter connectivity needs today. Beyond that, though, connectivity products today whether they are GbE or 10GbE must support the performance features of the evolving datacenter technologies. This includes: Scaling on multi-core processor-based servers Supporting the I/O arbitration needs of multiple virtual machines (VMs) in sever consolidation and virtualization Internet Small Computer System Interface (iSCSI) support for storage-over-Ethernet applications

With 1Gb networking the defacto standard in datacenter environments, getting companies to migrate to a networking technology to replace their existing environment is a hard pill to swallow. However, we believe the ramp for mass adoption of 10Gb Ethernet which started over the past couple of years could accelerate significantly with the broader adoption of 10Gb connectivity on the server motherboard (LOM) on servers shipping with the Romley upgrade later this year. We strongly believe the inclusion of the 10Gb NIC LOMs is going to be the driving force that pushes adoption of the 10Gb Ethernet into the datacenter. Our belief is based on the fundamental premise that businesses are hesitant to purchase new infrastructure when there is an incremental expense. However, when the new infrastructure is included with the new purchases they make, suddenly the investment becomes less and they make purchases to leverage the new technology they possess. The simple facts that the 10Gb Ethernet ports are included on their servers and are backwards compatible with the standard 1Gb Ethernet deployed in their environment, we believe could help customers finally adopt this technology in wider numbers. Exhibit 11: 10GbE in a Typical SAN Environment

Source: Intel, Inc.

Page 16

March 14, 2012

Industry Report

Storage companies are also not waiting. A majority of the private companies showcased in this paper, assuming that 10Gb Ethernet will be the predominant storage interconnect in the future, are designing their equipment to leverage this trend. Although 10GbE has been around for several years, we believe, it is now poised to take a position as an alternative fabric for SAN applications. This was made possible by the Internet Small Computer System Interface (iSCSI) standard. The iSCSI standard is an extension of the SCSI protocol for block transfers used in most storage devices and also used by Fibre Channel. The Internet extension defines protocols for extending block transfer protocol over IP, allowing standard Ethernet infrastructure elements to be used as a SAN fabric. Basic iSCSI capability is implemented through native iSCSI initiators provided in most OSs today. This allows any Ethernet NIC to be used as a SAN interface device. However, lack of a remote-boot capability has in the past left such implementations lacking in the full capabilities provided by Fibre Channel fabrics. Initially, iSCSI host bus adapters (HBAs) offered a solution, but they were expensive specialty items much like Fibre Channel adapters. To resolve the remote-boot issue iSCSI Remote Boot support is now available with PCIe server adapters, including the new generation of 10 Gigabit Ethernet Server Adapters. This allows use of the greater bandwidth of 10GbE in new SAN implementations. Additionally, Ethernet and Fibre Channel can coexist on the same network. This enables use of the greater performance and economy of 10GbE in expanding legacy Fibre Channel SANs as shown in Exhibit 11 above. We also note the use of new and simpler Ethernet-based storage protocols such as the use of ATA over Ethernet (AOE) by Coraid has become possible with the use of 10Gb Ethernet and potentially driven the economics of enterprise storage down to a whole new level which was hitherto impossible to attain in the world of Fibre Channel. Exhibit 12: 10GbE Connectivity Innovation Using ATA Over Ethernet

Source: Coraid, Inc.

3.4

Server, Storage and Desktop Virtualization

The industry reached a virtual tipping point in 2009 when, according to IDC, the number of newly installed virtual machines surpassed the number of newly installed physical servers. This inflection point is having a profound impact on how IT resources especially Storage are managed, secured and provisioned. It is not surprising that server virtualization has so quickly penetrated enterprise data centers. The technology has a profound impact on organizations CapEx and OpEx. CapEx (higher utilization means fewer servers need to be purchased) and OpEx (few servers means a simpler

Page 17

March 14, 2012

Industry Report

environment and lower maintenance costs) can be reduced. According to IDC, virtualization can also save 83%of floor space (which costs upwards of $1,000/sq.ft) and 87% of energy costs. Although blade servers tend to increase density within the data center, virtual machines have a far more profound effect as they can be created as required with little additional "cost". Besides increasing the compute density on physical servers, server virtualization has also enabled the pooling of compute capacity across several physical machines. Please see Exhibit 13. Technologies such as vMotion have driven the adoption of network storage in environments where IT has never used Network storage in the past. Exhibit 13: Pooling of Server and Storage Capacity Enabled by Virtualization

Source: ThinkEquity LLC Similarly, on the storage side, virtualization has enabled the pooling of storage across physical storage appliances and thus enabled IT to reap benefits of better storage capacity utilization, flexibility of provisioning storage to applications and its management in response to variable demand. Lately, solutions from some vendors have also enabled the pooling of storage capacity across systems from different vendors (heterogeneous storage virtualization), which has in turn, also enabled the aforementioned benefits to be further enhanced. Storage can now be managed as an abstract capacity, indifferent from the underlying location of physical disk. Although virtualization has led to a significant increase in compute density, storage has not kept pace. Denser server workloads mean significantly more and diverse traffic between servers and storage traversing the pipe. Storage arrays now have to face more random I/O as well as a very diverse set of workloads in virtual server environments. Storage virtualization, a technology that enables pooling of diverse storage systems is an answer, but we believe a more radical design of storage controllers is probably a longer term answer which companies profiled in this paper are attempting to do. 3.5 Virtual Desktop Infrastructure (VDI)

Virtual desktop infrastructure (VDI) refers to the emerging practice of hosting a desktop operating system within a virtual machine (VM) running on a hosted, centralized or remote server. Many enterprise-level implementations of this technology store the resulting "virtualized" desktop on a remote central server, instead of on the local storage of a remote client; thus, when users work from their local machine, all of the programs, applications, processes, and data used are kept on the server and run centrally. This allows users to run operating system and execute applications from a smartphone or thin client which exceed the user hardware's ability to run. The shared resources model inherent in desktop virtualization offers advantages over the traditional model, in which every computer operates as a completely self-

Page 18

March 14, 2012

Industry Report

contained unit with its own operating system, peripherals, and application programs. Overall hardware expenses may diminish as users can share resources allocated to them on an as-needed basis. Virtualization potentially improves the data integrity of user information because all data can be maintained and backed-up in the data center. Other potential advantages include simpler provisioning of new desktops, reduced downtime in the event of server or client hardwarefailures, lower cost of deploying new applications, desktop image-management capabilities, longer refresh cycle for client desktop infrastructure and secure remote access to an enterprise desktop environment. Exhibit 14: Typical Desktop Virtualization Deployment

Source: VMware, Inc. The impact of VDI on storage requirements is significant and has been a big driver of innovation in storage. If not managed smartly, storage capacity needs for desktop virtualization can expand dramatically as VDI administrators clone thousands of desktops and can destroy the economics of a VDI initiative. According to Atlantis Computing, Storage Capex on a VDI initiative can represent up to 80% of the total budget. This requires a storage system that can support snapshots, clones and thin provisioning for the hundreds or thousands of virtual desktop images a large VDI environment can have. These storage systems must have the performance to support very high numbers of snapshots and clones, data efficiency to keep storage capacity under control and a scalable, file-based architecture to meet the challenges posed by desktop virtualization. Data protection for VDI can be specifically tricky especially when it comes to restores. However, the biggest challenge posed to VDI storage in our view is in terms of performance. In terms of performance, simultaneous boot-up of several desktops can overwhelm a backup storage system. During normal operation, VDI can overwhelm the latency of a disk based system with the random I/O associated with several desktop users working simultaneously off a centralized system. There are also virus scanning and OS upgrade operations that occur from time to time, which triggers a spike in writes. The root cause of the VDI storage performance problem is that the Microsoft Windows Operating System is designed for a dedicated low latency PC hard drive and is extremely inefficient when that hard drive is replaced with a high latency storage system shared by many virtual desktops. As a result, to achieve the equivalent performance of a physical PC, VDI

Page 19

March 14, 2012

Industry Report

requires an enormous amount of storage to support the many I/O intensive tasks required by Windows storage that is used to service OS and application requests rather than store user data. More than simply adding capacity, scaling a VDI storage system requires scaling horsepower as well, to support the randomness thats typical in these environments and to accommodate the performance spikes. Exhibit 15: I/O Profile in a Typical VDI Storage Environment

Source: Nimble Storage The virtual desktop infrastructure creates unique challenges for storage systems, especially at the scale required to make these projects economical. The performance, capacity and management burden VDIs generate can outpace many storage solutions not specifically designed to meet these unique requirements. Scalable, high-performance, file-based storage systems are in our view, key components in a successful VDI implementation and major contributors to an acceptable ROI.

Page 20

March 14, 2012

Industry Report

3.6

The Cloud

We view the Cloud as the more virtualized and more automated next iteration of the data center. The Holy Grail for the Cloud is the notion of making all data center resources available as a service on-demand by abstracting them from the underlying hardware primarily using virtualization. Key to this notion is on-the-fly provisioning of compute and storage capacity and the flexibility of management that comes with these resources being abstracted from the underlying hardware. Agility and flexibility to rapidly changing business requirements have been over-arching goals for IT since time immemorial. Rapid provisioning of IT infrastructure in response to rapidly changing business requirements and variable workloads provides IT with the agility and flexibility it has yearned for in the past. In theory, cloud computing creates the ultimate in data center flexibility. If an application needs more resources, it can be added on the fly. If a workload needs to be moved to a higher performing server because business policy dictates it, that can be done in real time. Exhibit 16: Typical Cloud Deployment

Source: VMware, Inc. The Cloud is also spawning several business models as far as storage is concerned. Amazon S3, Google Storage and Azure from Microsoft are well-known Hosted or Public Cloud storage platforms targeted at the consumer and the SMB/SME market. We also believe the enterprise could see a significant part of its non-core workload move to Cloud storage models public, private and hybrid. Cloud storage in the enterprise has been off to a slow start mainly due to concerns over data security and also because the market pioneers did not do a great job of positioning the value proposition to the enterprise, in our view. Lately, concerns over security have taken a back seat as IT customers have

Page 21

March 14, 2012

Industry Report

realized that Public Cloud vendors are likely to do a better job at security considering their business model depends on it. No one has been more vocal about this dynamic than former US Federal CIO Vivek Kundra who proclaimed that the only issue with Cloud security was job security of internal IT. Part of the reason why Cloud storage has gained in popularity in the enterprise is that most CIOs have no choice as they struggle to manage rampant data growth in the face of a shrinking budget. Managing storage and data migrations from one platform to another especially for non-mission critical backup and archive data is not only time-consuming but far more expensive than the value of the data itself. A pay-as-you-go model does have compelling appeal in this scenario. In this paper, we feature an interview with the CEO of enterprise Cloud storage vendor Nirvanix. According to Scott Genereux of Nirvanix, the key trend that is driving Cloud storage growth is the rapid growth of unstructured data in the enterprise and the inability of existing on-premise storage solutions such as nearline archiving and tape to satisfy the speed, cost and compliance demands of enterprises simultaneously. Nirvanix is one vendor that offers all three flavors of Cloud storage public, private and hybrid. We believe the value proposition of a Cloud storage service needs to encompass the following features: Usage-based pricing based on actual consumption Elasticity to rapidly scale both up AND down Multi-tenancy allowing for businesses and departments to be managed separately Aggregation of storage resources into a single, universally accessible pool A Web-based object store with support for cloud APIs (REST & SOAP) A service rather than hardware or just software, with accompanying SLAs

Exhibit 17: Hybrid Cloud Storage

Source: Nirvanix We also profile StorSimple in this white paper which in effect has built an Enterprise Cloud Storage gateway appliance. The StorSimple solution takes the form of a hardware appliance and proprietary software that sits between applications systems and public or private storage clouds and performs a variety of operations on live data which are transparent to both the applications and the cloud (Exhibit 18 below). According to Ursheet Parikh, CEO of StorSimple, the company is in the process of delivering the iCloud (Apples Cloud service for consumers) for the enterprise. StorSimple provides cloud integrated enterprise storage that combines the function of primary, backup, archival and disaster recovery into a single appliance that is seamlessly integrated into cloud storage services such as Amazon S3 and Windows Azure. The companys core value proposition is the ability to combine primary backup, archival disaster recovery and tape elimination in a single package with a single device that seamlessly integrates into the cloud and allows the customer to use the cloud for disaster recovery. StorSimples key innovation is a concept called the cloud snapshot. Cloud snapshot fundamentally

Page 22

March 14, 2012

Industry Report

drives cloud data portability, across cloud providers. During a disaster, the speed with which you can recover in the context of a cloud snapshot is several orders of magnitude better than what traditional on-premise approaches offer in terms of recovery time when it came to tape, or in terms of cost compared to a traditional replication model for storage. Please read CEO Ursheet Parikhs interview in this paper for further detail on this company. Exhibit 18: Enabling Cloud Storage Leveraging a Cloud Gateway Appliance

Source: StorSimple

Page 23

March 14, 2012

Industry Report

4.0

Storage Myths Being Busted!

Page 24

March 14, 2012

Industry Report

4.1

Enterprise Storage is Network Storage

This has been the defining characteristic of the Storage industry over the past twenty years with the advent of SAN and NAS in the early 1990s. Enterprises have embraced the use of shared storage on the network as a means of efficiently and securely managing data in a multi-server environment. However, in the past few years, we have seen several new companies come ahead and challenge this status quo. No other company has been as successful as Fusion-io in moving persistent Storage back to the server in the enterprise, in our view. However, we see several other models that seek to meld computing and storage together in order to reduce latency between compute and storage or to make data processing faster as storage capacities swell to petabyte scale. Amongst companies featured in this paper, Nutanix has a converged compute and storage architecture which leverages a scale-out distributed file system. The argument of converging compute and storage in a single box at Nutanix is to reduce latency between compute and storage. NexGen Storage leverages a Fusion-io PCIe card on the server for the same reason as Fusion-io because it believes the best place to leverage the performance of Flash storage is right on the compute bus given the lack of any layers that could bring in latency otherwise. All-Flash storage array vendors such as Violin and Kaminario also leverage PCIe connections to connect directly to the compute bus and reduce latency that networking equipment such as switches and adaptors might bring. In the Big Data market, the ability to process petabyte scale data stores efficiently and at speed is at a premium. Data Direct Network, a private company that we profile in this paper, has trademarked the concept of in-storage computing to define a concept of moving silicon based processing power right next to the data store in order to get an unprecedented level of performance out of raw disk. The companys compute technology on the storage controller parallelizes data transfer and protection through hardware acceleration to deliver optimized large file read and write performance that would otherwise not be possible. This performance is extremely useful in big-data workloads such as high-performance computing research, media and entertainment, genomic sequencing, media and entertainment and oil and gas exploration. Please read DDN EVP Jean-Luc Chatelains interview later in this report. Exhibit 19: In-Storage Computing Architecture

Source: DataDirect Networks

Page 25

March 14, 2012

Industry Report

4.2

In storage, you have to choose between performance and capacity

Traditionally, storage systems have focused on optimizing either capacity or performance. Primary storage systems were typically optimized for performance while backup or secondary storage systems were typically optimized for capacity. However, a new breed of private companies is working hard to upend this notion. A common thread runs across solutions of all these companies they all leverage in-line de-duplication and flash in their primary storage solutions. Nimble Storage offers a converged primary and backup appliance that leverages inline de-duplication and compression to write random I/O to a Flash storage tier. The random I/O is then converted into serial I/O for persistent storage on disk which also includes a tier for backup storage. Up to 90 days of backup can be retained on single array which also performs as a primary storage array. Since there is no need to move data to or from backup media, backups and restores can be performed in seconds. Pure Storages solution also performs in-line de-duplication on primary storage data that then gets stored on MLC-based SSDs in all Flash primary storage array. According to Pure this combination of de-duplication and MLC-based SSDs results in both better performance and cost per Gb compared to enterprise-grade, spinning disk-based systems. Nimbus Data Systems has a similar strategy as Pure Storage and essentially upends conventional storage thinking in a way that has not been attempted before. 4.3 In storage you provision capacity, not performance

Traditionally storage capacity has been provisioned to applications and Hosts by carving out LUNs in shared storage environments. When it came to provisioning performance, storage administrators were limited in their ability and the only degree of performance provisioning they could do was at a high level such as enabling short-stroking of disks to drive performance. Two companies profiled in this white paper NexGen Storage and SolidFire have developed technology that enables granular level QoS-based performance provisioning to Applications and Hosts. Both companies technology enables performance provisioning at the virtual server level. NexGen is targeting the Enterprise market vs. SolidFires focus is on the Cloud Service provider market. We believe this feature addresses an unmet need in the enterprise and Cloud datacenter respectively. 4.4 Enterprise grade Block Primary Storage will always be expensive

Coraid with its connectionless bare metal ATA over Ethernet-based SAN, we believe is turning this notion on its head. This technology compared to a traditional SAN does not use Fibre Channel as a storage interconnect. It does not use expensive network switches or adaptors and leverages an open-source initiator that is embedded within the Linux kernel to connect hosts to storage in a massively parallel fashion that not only offers lower latency than a typical Fibre Channel connection but economics that are close to a fifth of a typical Fibre Channel based SAN. Coraid offers its Ethernet-based SAN storage at prices that start at $0.50 per Gb of capacity which we believe comprehensively upend the notion that SAN is typically expensive storage.

Page 26

March 14, 2012

Industry Report

5.0

New Storage Product Markets Being Spawned

Page 27

March 14, 2012

Industry Report

Product Market Snapshot

Product

Market

HPC

Enterprise

Web 2.0/Cloud

Server Side Flash

All-Flash Arrays

Hybrid Arrays

Ethernet SAN

Converged Infrastructure

Storage Optimization Appliances

Sources: Respective companies websites

Page 28

March 14, 2012

Industry Report

5.1

Server-Side Flash

NAND Flash-based DAS or server-side Flash storage has in our view, emerged as a very real solution to the server I/O bottleneck. Aside from the obvious drawback that plagues any DAS solution in terms of inability to share storage across several hosts, we believe the best place to leverage the speed and low-latency of Flash storage is the CPU bus. We believe the closer fast storage resides to processors, the better the performance of the system. We believe Server-side Flash storage has turned traditional thinking related to enterprise storage on its head by creating a new tier of memory based on NAND Flash, which emulates block-based network storage. This new tier of memory is persistent storage. It resides close to the processor, is higher capacity than traditional DRAM-based memory and orders-of-magnitudes faster in I/O performance compared to spinning disk based network storage. As discussed earlier in this paper, spinning disk is no longer seeing an improvement in I/O speeds and latency while Moores Law has ensured that semi-conductor based NAND flash storage has continued to improve in I/O speeds and reads over the past decade. Hence, as result, NAND flash based storage today offers several orders-of-magnitude better I/O than the best spinning disks. Besides, the data center network in its traditional and currently most prevalent multilayered hierarchical form, causes latency. This further reduces the I/O possible with spinning disk on network storage. Hence, a combination of server-side flash based DAS, which can be connected to a server using a native PCIe connection, becomes a very useful solution in the battle to eliminate the server I/O bottleneck and reduce network latency. We illustrate a typical deployment configuration in the Enterprise where a PCIe-based card could be used in conjunction with HDD storage. A key enabler of server-side Flash storage has been the advent and growing popularity of open-source Distributed File System (DFS) to manage scale-out server deployments. Data centers are no longer constrained by software licenses they deploy as they scale out their servers in the quest for more compute capacity. Commodity servers using these Distributed File systems can be scaled on demand. Storage on the server makes a lot more sense in this scale-out mode given the relatively lower complexity and cost compared to a scenario where network storage and switches have to be configured and provisioned for each server. The storage meta data gets handled by the DFS and thus the scale-out deployment knows where specific data is stored on flash or disk storage on the server. Exhibit 20: A typical datacenter deployment leveraging server-side flash and HDD-based network storage

Source: SNIA

Page 29

March 14, 2012

Industry Report

As an illustration of this value proposition, we highlight a Fusion-io case study where flash-based DAS was used to replace a SAN based solution in a data warehouse application setting utilizing a cluster of servers. Fusion-ios customer saw a 4x decrease in data access latency and a 40x decline in query processing time for the data warehouse application, with the introduction of Fusion-ios flash-based DAS solution. Exhibit 21: Fusion-io Case Study Illustrating Impact of Flash-Based DAS on Server I/O

Source: Fusion-io

Source: Fusion-io

Page 30

March 14, 2012

Industry Report

The specific benefits of server-side Flash storage can be summarized as follows: Significantly higher throughput and lower latency: As illustrated above using a Flash-based DAS solution enables servers to achieve significantly higher throughput and lower latency and virtually eliminate the server I/O bottleneck problem. Lower Costs: Better server I/O leads to better CPU utilization. It reduces the number of servers and switch ports required as more workload can be consolidated on fewer servers, in our view. Energy Efficiency: By reducing the amount of hardware required and hence floor space, this solution in our view can also significantly reduce energy consumption in a data center.

As transistor density in flash memory keeps improving, and hence the storage density of flash memory keeps increasing, the capacity of flash storage devices that can be attached to a server will also keep increasing in the future. Currently, Fusion-ios largest storage product, ioDrive Octal, packs about 10.24Tb of capacity, which is less than what can be made available in large network storage arrays. However, as Moores Law takes effect over time, this problem will diminish in the future, in our view. The Players Fusion-io appears to have taken the early lead in marketing this solution and today is going-to-market through the topthree server OEMs in the market IBM, HPQ and Dell. EMC with the launch of its Project Lightning product VFCache has also entered the fray in this promising product market. Texas Memory Systems (TMS), STEC, OCZ, Micron and LSI are other companies that may participate in this market in the future. We recommend reading TMS President Dan Scheels interview in this white paper to not only learn about the innovation at TMS but also about this very important product market. However, we wish to point out that all PCIe cards in the market today are not the same in terms of architecture and the resulting performance. A key difference in architecture between Fusion-io and the other vendors is in terms of where the flash management, RAID management and other management software operates. In the case of Fusion-io, this software resides on the server operating system whereas in the case of the other vendors, these functions are enabled by firmware resident on specific chips (controller ASICs/FPGAs) within the card. This is a key difference and in our view gives Fusionio with lower latency characteristics although it does consume some CPU cycle on the host. Three private companies profiled in this paper Kaminario, NexGen Storage and Nutanix leverage Fusion-io solutions and not other competing and potentially cheaper solutions available in the market. Please refer to Exhibit 22 below to understand the key differences in architecture between FIOs ioMemory and competing PCIe solutions and SSDs residing inside the network storage array.

Page 31

March 14, 2012

Industry Report

Exhibit 22: Comparison of FIOs architecture with competing traditional network storage SSD and PCIe SSD form factor-based solutions Traditional Network Storage based SSD data flow: data traverses nine layers before hitting the SSD flash memory.

Data flow with a PCIe SSD: data traverses five layers before hitting the flash memory on the SSD

Data flow with a FIO card: data traverses only two layers before hitting the flash memory.

Source: Fusion-io

Page 32

March 14, 2012

Industry Report

5.2

All-Flash Storage Arrays

Several private companies at this point of time are leveraging Flash inside a network storage array as the only storage medium. Interestingly, no two companies profiled or interviewed in this paper utilize the exact same strategy and architecture. A key commonality across all these players is that they have created a new type of storage architecture specifically designed for SSD media. We do see some similarities in terms of market focus and utilize it to segment what we view as this exciting new product-market category as follows: Exhibit 23: All-Flash Storage Array Market Segmentation All-Flash storage array segments based on market focus High-performance Database Applications and Analytics Enterprise Participants Kaminario, Texas Memory Systems, Violin, Project Thunder (EMC), Huawei Symantec Dorado Nimbus, Pure Storage, WhipTail

Cloud

SolidFire

Source: ThinkEquity LLC High-Performance All Storage Array

A primary focus of companies in this segment is to leverage the raw performance of flash in terms of throughput and I/O to enable high-performance database reads and writes in applications such as data warehousing, reporting analytics, highend OLTP, search-index, high-frequency trading, virtual server and virtual desktops. Companies in this segment have essentially focused on developing an architecture that can leverage the performance of mainly raw Flash and overcome the data supply problem caused by the latency and throughput stagnation that plagues HDD today. Although these systems have some basic data management software included with them, they essentially focus on providing fast block storage in terms of throughput and low latency and leave the task of data management to file system or volume controller software that could be placed between the storage system and the host. Typically, a lot of traditional storage systems function such as RAID, page mapping and garbage collection is conducted in hardware and not in software to minimize latency. Although these all flash storage arrays can be placed on a network storage location, they can also be connected to a host using a PCIe connection, as evident in Violins switched memory architecture depicted in Exhibit 24 below. The switched memory architecture enables greater scalability and fault tolerance compared to memory bus architecture.

Page 33

March 14, 2012

Industry Report

Exhibit 24: Architecture of an All-Flash Storage System using memory modules

Source: Violin Memory

There are some differences in how Flash is used in the storage arrays across vendors in this segment. Violin utilizes Flash in a memory module format as evident from the Exhibit above. Violin calls the memory module a VIMM. According to Violin, using raw Flash instead of SSDs bestows its systems with extra-ordinary low latency considering there are no layers between the flash controller and the raw Flash. Kaminario, on the other hand, leverages DRAM as a tier of lowerlatency but also leverages Fusion-ios PCIe-based Flash cards see Exhibit 25 below. TMS utilizes its own Flash PCIe cards within its all flash storage arrays. Exhibit 25: All-Flash Storage Array Leveraging Flash and DRAM

Source: Kaminario

Page 34

March 14, 2012

Industry Report

Enterprise-Focused All-Flash Arrays

Vendors in this market segment in our view essentially see Flash based systems as a replacement for the traditional HDD-based enterprise network storage market. The focus of this all-flash storage segment is to bring the performance of flash to the enterprise at the economy of HDD-based storage while providing all the software based data management features that are table-stakes in the mainstream enterprise network storage market. Besides three vendors that we have profiled and interviewed in this paper Nimbus, Pure Storage and WhipTail other vendors targeting this segment include XtremeIO. While the solutions differ in their respective implementations, a common element across the products of companies in this segment is that they all offer a full suite of data management and data protection technology designed from the ground up for flash based storage. It is this software capability that separates these vendors from the highperformance segment of all-flash storage arrays. Typical enterprise workloads where products in this segment find favor include virtual server and desktops, e-mail, and typical transactional database environments which need block storage. Nimbus differentiates itself from the other players in this segment with the fact that it has developed its own hardware. The company has developed its own flash modules that Nimbus believes have higher endurance, usable capacity and speed than anything similar in the market. Nimbus' HALO operating system is a comprehensive software suite that provides endto-end administration, data protection, data optimization, data security, and detailed monitoring of Nimbus systems. Please note the comprehensiveness of the software suite and protocol support compared to the architectures discussed in the previous section. For a more detailed deep-dive into Nimbus and its strategy, please read the complete interview with Nimbus CEO Tom Isakovich later in this report. Exhibit 26: Comprehensive Enterprise-grade Flash-Optimized Storage Software Suite

Source: Nimbus Data

Page 35

March 14, 2012

Industry Report

Pure Storage, on the other hand differentiates itself on being able to offer enterprise grade, software-optimized, all-Flash storage at a price lower than the price of a performance-disk based system. The company is able to achieve this by leveraging global, in-line de-duplication and compression to enable 5-20x data reduction while using MLC-based standard SATA interface SSDs for sub-millisecond latency. According to Pure storage, the key intellectual property in the companys solution is the ability to deliver effective data deduplication and with sub millisecond latency. This is baked into the companys comprehensive software suite which leverages its Purity operating environment. See Exhibit 27 below. The software also optimizes the reliability of Flash memory within the SSDs. According to Pure, it can take Flash that has been written well past its warranty and whose performance had degraded commensurately to provide performance that is on par with an absolutely new drive. Other enterprise features that Pure currently delivers in its arrays include hardware redundancy, self-healing, snapshots and replication see Exhibit 27 below. For a more detailed deep-dive into Pure Storage and its strategy, please read the complete interview with Pure Storage CEO Scott Dietzen later in this report. Exhibit 27: Comprehensive Enterprise-grade, Flash-Optimized Storage Software Suite

Source: Pure Storage

The third company that we have profiled in this all-Flash enterprise-focused storage array segment is WhipTail, which is led by ex-Brocade CTO Dan Crain. The company has purpose-built its storage systems to behave like a 100% solid-state storage arrays based around NAND Flash technology. The companys IP leverages NAND flash to provide very high symmetrical data-writing and read performance and to extend the life of NAND Flash to bring it on par with mechanical disk. The companys initial success has been with random I/O environments such as VDI and virtual server environments and in database intensive Oracle/SQL environments. For a more detailed deep-dive into WhipTail and its strategy, please read the complete interview with CEO Dan Crain later in this report.

Page 36

March 14, 2012

Industry Report

All-Flash Cloud-Optimized Storage Arrays

SolidFire distinguishes itself from enterprise-focused players with its single-minded focus on the needs of Cloud Service providers. The companys all-Flash storage arrays come equipped with the SolidFire Element operating system which we believe includes most enterprise-grade software features such as thin provisioning, compression and de-duplication, cloning and self-healing but also incorporates Cloud specific features such as scale-out to 100 nodes, multi-tenancy and Quality of Service provisioning that make its All-Flash storage arrays particularly well-suited for the growing Cloud service provider market. We view the Quality of Service provisioning function very unique (only NexGen Storage amongst the companies we spoke to has this function) and we believe it gives cloud service providers the ability to set and guarantee storage performance for every customer. For a more detailed deep-dive into Nimbus and its strategy, please read the complete interview with SolidFire CEO David Wright later in this report.

Exhibit 28: Flash-based Cloud-optimized Storage

Source: SolidFire

Page 37

March 14, 2012

Industry Report

5.3

Flash-Optimized Hybrid Storage Arrays

Within this segment of Next-generation Storage, we identify two companies that are utilizing Flash in interesting ways to optimize their storage system while leveraging more economical SATA drives for persistent long-term storage. Both NexGen and Nimble Storage are targeting the iSCSI block storage market. They both use automated data tiering in innovative ways along with other unique architecture and software based features to deliver their respective value propositions. The key value proposition for NexGen Storage is control of application performance (Quality of Service) in a virtual-server, shared-storage environment. The company has attempted to address the problem of controlling performance of an application running inside a virtual machine on a server via what we see is essentially dynamic allocation of an appropriate storage tier back to specific virtual machines. We think this feature is particularly unique amongst all enterprise-centric storage companies with the exception of SolidFire which appears to be doing something similar but with a focus on the Cloud service provider market. NexGen, which is founded by the team that started LeftHand Networks, leverages Flash as a tier of storage right on the compute bus via a Fusion-io PCIe card. The company sees itself as a shared Fusion-io solution that can be shared across multiple servers. As seen in Exhibit 29 below, NexGen also leverages cheaper SAS disk for lower performance economical storage. The companys technology also leverages the greater bandwidth made possible with 10Gb Ethernet and PCIe on the compute bus, besides flash and multi-core processors. Please read CEO John Spiers interview in this paper for further detail. Exhibit 29: Flash-Optimized Hybrid Storage Approach at NexGen Storage

Source: NexGen Storage At Nimble Storage, the team, comprised of several Data Domain and NetApp alum, has attempted to develop the most versatile storage system in the SMB market today leveraging the latest technology advances available to the storage

Page 38

March 14, 2012

Industry Report

industry today and in the process converging primary, backup and disaster recovery into one storage solution. The companys iSCSI block-storage solution blends the benefits of Flash and high-capacity disk in terms of performance and capacity respectively via its unique Cache Accelerated Sequential Layout (CASL) file system. Hot data (see Exhibit 30 below) is cached on SSDs while 100% of data is written to cheaper higher-capacity disk which sits behind the disk. Random I/O gets handled by the Flash layer initially and then the CASL converts the random I/O to serial I/O to write the data to disk, thus leveraging the Flash layer for performance and the spinning disk for capacity and economics. We also note that using Flash for tier0 of storage enables the system to be used for a wide variety of applications including those demanding random I/O. The CASL also compresses all data in-line which further boosts the economics of the system, in our view. Nimble also adds another unique capability in terms of providing not just primary storage but also backup storage off of one system. According to Nimble (please read CEO Suresh Vasudevans enlightening interview in this paper), the company has taken all of Data Domains approaches of de-duplication and inline compression in backup storage and instead of building those into a separate storage system, Nimble has it on the same system where application data or primary storage is being housed. This combination of Flash-based performance and disk based economics, primary and backup in the same box, appears fairly unique and may prove to be very compelling to the SMB market which currently is served by a somewhat less versatile set of products such as Dell EqualLogic, HP LeftHand and EMC VNXe. Exhibit 30: Flash-Optimized Hybrid Storage Approach at Nimble Storage

Source: Nimble Storage

Page 39

March 14, 2012

Industry Report

At Tintri, which incidentally means Lightning in Gaelic, a team of ex-VMware, Data Domain, Sun and NetApp executives has essentially tried to fill the gap in the traditional network storage market where most arrays were built for the physical world (not virtual machines) and hard disk drives, with a purpose-built storage appliance targeted at the virtual machine world, leveraging Flash as the primary storage medium. As is evident from Exhibit 31 below, Tintri has attempted to eliminate a lot of arcane technology concepts associated with the physical and HDD world such as LUNs, volumes and zoning, with a system that is simple to manage virtual machines with and is built specifically for the virtual environment and for Flash SSDs. Its a combination of two things. It is a storage system that is built exclusively for virtual machines, and its also one that is built for Flash. This combination allows Tintri to eliminate the performance hit associated with the virtualization of database intensive applications. Tintri claims that 99% of all IOPS in its storage systems comes directly from Flash. This makes the system very suited for the random IOPS associated with the virtual world both server and desktop virtualization use cases, in our view. The company also leverages de-duplication and compression and disk behind the Flash SSDs to keep the solution at a price point which is below those of traditional network storage appliances. Please read CEO Kieran Hartys interview in this paper for further detail.

Exhibit 31: Storage Purpose-Built for Virtual Machines and Flash

Source: Tintri

Page 40

March 14, 2012

Industry Report

5.4

Converged Infrastructure (Leveraging Flash Storage)

In this segment of the market, Nutanix has developed a scale-out converged compute and storage infrastructure solution targeted at virtualized datacenters. Nutanix Complete Cluster is a scale-out cluster of high-performance nodes, or servers, each running a standard hypervisor and that contains processors, memory and local storage, including SSDs and hard disk drives. Each node runs virtual machines just like a standard virtual machine host. In addition, local storage from all nodes is virtualized into a unified pool by Nutanix Scaleout Converged Storage (SOCS). The secret sauce for Nutanix is basically a Google file system like architecture that is hosted inside the hypervisor. The company has made the storage controller fully virtual, an arrangement in which storage servers sit next to the application servers on the same hardware. Nutanix believes (please read CEO Dheeraj Pandeys interview later in this paper) that there is a massive opportunity to share the motherboard, the fans, the power controllers, the chassis, the network, the management console, and the manpower to achieve integrated converged efficiencies between tiers of the data center that were hitherto territorialized. Nutanix's key innovation is the company's patent-pending distributed system software that provides a converged and scale-out storage layer, is designed specifically for virtualization, and is optimized from the ground up to make use of Flash SSDs in its core architecture. This converged architecture approach, pioneered by Google and built-upon by Nutanix, envisions datacenters that are much smaller and faster than those with SAN-based architectures, yet simpler and more cost-effective to set up, scale, and sustain. Each Nutanix node (a building block of a cluster) contains six Intel cores worth of compute capacity, 320GB of Fusion-io PCIe SSDs, 300GB of SATA interface SSDs and 5TB of raw SATA HDD capacity. Four nodes can be clustered together to make a Block. In essence, the Nutanix converged architecture reduces system latency by eliminating the network and by using PCIe SSD cards from Fusion-io close to the compute bus, while at the same time maintaining storage as a shared resource through the use of a scale-out distributed file system. While we view this architecture as revolutionary, we do see it as hugely disruptive if it succeeds in the enterprise to any extent as much as it has in the Web 2.0 space, considering the performance advantages it brings along with the simplicity of management.

Exhibit 32: Scale-Out Converged Storage

Source: Nutanix

Page 41

March 14, 2012

Industry Report

5.5

Ethernet SAN

While Ethernet SANs using iSCSI have existed for over a decade and represent a long-established product market, Coraid has pioneered the use of ATA over Ethernet (AoE) as an interconnect which essentially enables Fibre Channel like or better performance over Ethernet at a fraction of the cost. Coraid essentially leverages the advances and improving economics in 10Gb Ethernet, multi-core processor based servers and commodity disk in its solution to provide multi-core processors running Linux massively parallel access to storage capacity without the use of expensive adaptors and switches. The solution utilizes the ATA over Ethernet initiator inside the Linux kernel to connect servers to storage. The data has to traverse through very few layers to get to Storage, which enables low latency and superior performance. The Ethernet SAN is built on bare metal layer-2 Ethernet and as a result it makes the entire SAN look like an internal drive to each server. Coraids second layer of innovation is at the storage operating system layer. The companys CorOS is a distributed operating system, which enables scale-out storage that can scale to petabytes all connected in a massively parallel fashion to the hosts without traditional performance bottlenecks that plague storage systems connected through networking equipment like adaptors and switches. The combination of a connectionless, flexible network layer with a scale-out operating system gives customers the ability to start small and expand a system to handle large data growth. This combination may also enable unprecedented economics of a SAN with Coraid quoting prices as low as $500 per TB of capacity. The company also has a Flash-based solution that leverages the same architecture for SAN economics as Coraids disk-based solution but is built for a higher level of I/O performance.

Exhibit 33: Comparison of SAN Protocols

Source: Coraid

Page 42

March 14, 2012

Industry Report

5.6

Flash-Accelerated Storage Optimization Appliances

This product market segment focuses on optimizing performance and capacity of installed Storage systems by essentially leveraging Flash and tiering software. The value proposition of a storage optimization appliance is in terms of optimizing the performance or scalability of pre-existing disparate storage systems by essentially providing a performance offload to the main archive of data and not in terms of providing raw storage capacity. Players with products in this market segment include Avere Systems (please read interview with CEO Ronald Bianchini in this report), GridIron, Texas Memory Systems and CacheIQ. Typical workloads where storage optimization appliances have been seeing traction include high-performance NAS use cases such as media and entertainment, seismic and oil exploration where customers can use traditional NAS at the back-end and utilize an optimization appliance to accelerate I/O and performance. Another set of workloads where Avere has seen interest is branch offices accessing applications over the WAN from a central data center. An appliance can help accelerate traffic by providing a local cache at the branch office. Finally, a storage optimization appliance can potentially save on expensive high-speed networking in capacity intensive use cases by optimizing performance of cheaper disk based storage. Exhibit 34: Flash-Accelerated NAS Optimization

Source: Avere Systems

Page 43

March 14, 2012

Industry Report

6.0

Conclusions

We believe an exciting new wave of enterprise storage innovation is upon us driven by a new generation of promising startups that are leveraging emerging new technologies to solve growing problems, address new workloads and in the process bust several myths that have defined the storage industry until today. Like always, we expect competition from legacy storage vendors to heat up in the near-term. EMC, with its announcement of VFCache and Project Thunder, has signaled its intent to take on the first two product-market segments listed above. Hardly a day goes by before we hear a new product announcement from an existing large vendor or a private company. Ideas are coming out thick and fast, but we believe the real relative winners in this exciting dynamic are likely to be players that execute relentlessly and effectively not only in product development but also in go-to-market. That being said, we look forward to tracking this space closely and identifying the winners of tomorrows storage market.

Page 44

March 14, 2012

Industry Report

7.0

Interviews with Select Private Company Leaders

Page 45

March 14, 2012

Industry Report

Ron Bianchini, CEO and Co-Founder of Avere Systems Avere Systems provides NAS Optimization solutions designed specifically to scale performance and capacity separately and leverages Flash-based storage media using real-time tiering. Averes products allow organizations to achieve unlimited application performance scaling, free applications from the confines of the data center by eliminating latency and cut storage costs by more than half. As President and Chief Executive Officer of Avere Systems, Co-Founder Ron Bianchini has a long record of accomplishment in building and leading successful companies that deliver breakthrough technologies. Prior to Avere, Ron was a Senior Vice President at NetApp, where he served as the leader of the NetApp Pittsburgh Technology Center. Before NetApp, he was CEO and Co-Founder of Spinnaker Networks, which developed the Storage Grid architecture acquired by NetApp. Here are excerpts of his interview with ThinkEquity LLC: Rajesh Ghai (ThinkEquity): Ron, can you address the key problem youre trying to solve in the data center at Avere? Ron Bianchini (Avere Systems): What our products address is the high cost and poor performance of todays NAS (network-attached storage) installations. And, really, we call our product a NAS optimization product. What that means is we dont actually sell the archive storage the eventual archive storage of where your data is at rest. But we sell an appliance that sits in between the users and your archive solution and optimizes that solution primarily by providing offload from your archive store. Really there are three use cases that we go after. If you think of our appliance in the middle providing offload, the first use case is high performance computing. That encompasses financial modeling anything that requires a lot of I/O media and entertainment, seismic. We can scale to much higher performance levels than any of the incumbents in the NAS space. In this use case, we have a SPECsfs NFS benchmark posting where we are the highest-performing NAS filer, but also at a fraction of the cost and footprint of all the competitors. That is use case one. Use case two is distributed compute or a large, distributed enterprise. We hide the latency associated with WAN or cloudbased solutions. So the user thinks the file is near them even though the archive storage could be half a world or hundreds of milliseconds away. Then the third use case is cost, where we can provide enough offload from the archive filer, your eventual data store. Rather than using Fibre Channel or very high-performance and very expensive storage, you can migrate all of your storage over to very high-density, low-cost, low-power SATA-based storage. Moreover, you can get down to a fraction of your footprint, yet still keep performance and capacity the same. TE: Thats very interesting. And you mentioned that incumbents do not address this problem as well. Why do you think those incumbent vendors solutions do not address this problem? AS: The core technology inside a NAS solution is something called the file system. What weve done is created a file system that spans multiple media types. With all file systems up until Avere, theres one media type selected for the archive storage. You could pick that media type at the time you purchase the file system the NAS product. But from then on, that media type has to be the one that is used by that file system. What our file system, the internal technology, allows us to do is to span multiple different media types. So we use SATA (serial advanced technology attachment) for your archive data. We use SAS (serial attached SCSI) for your highperformance sequential read data. We use SSD (solid state drive) for your high-performance random data. And we deploy all of those media types under one single file system, which enables us to get both the high performance and the low cost at the same time. The incumbents and the traditional solutions in the market space to date have tried to do bolt-on fixes to address this problem. There really is no single file system that enables the use of all media types at once. TE: And what are key market trends that you see in terms of demand for your solutions? AS: I think that the key trends really map back to our use cases. In the performance use case, its all this additional processing. Its all database analysis. So you can think about financial modeling and then there are the traditional media

Page 46

March 14, 2012

Industry Report

and entertainment applications. All of the database work that even the corporate enterprise does is really driving the performance. More and more decisions are now data-oriented. And thats what drives performance. For the WAN-use case, its really the distributed enterprise. People now have employees all over the globe and give them access to a shared data pool, but giving them that access at high performance really is whats driving that use case. And the cost use case is simple. Its huge data growth and up until Avere, if you wanted to get both performance and capacity, you are typically buying very high performance but low density Fibre Channel drives. With us, you can buy the highest density SATA drives for the archive and you deploy Avere for performance. So its really the trends in those three use cases. TE: And when you look at those three use cases, what is the size of that addressable market that you think you can target for these solutions? AS: We target any part of the NAS market. The NAS market itself is a +$5 billion market and our solutions absolutely plug-in directly if you have a NAS installation. But the really interesting thing is because we can drive cost down so much drive performance well beyond where NAS used to be, I think our solutions are much more applicable to the general enterprise storage market which is much more like a +$30 billion market. So I think +$5 billion target just for the NAS space. But because of the new performance envelope we provide we think its much more applicable to the general enterprise storage market. TE: Thats very interesting. And if you look at your solution, what are the key technology enablers of your solution at this point in time? AS: The technology enabler is really our underlying file system technology. The fact is that solid-state storage has been moving more and more into the enterprise. And the problem without Avere is that you had to make a wholesale switch to that. Yet we allow you to get the performance of solid state with the capacity-scaling of SATA. So its really the new storage technologies that were leveraging and were using tiering algorithms to enable us to use all those different media types. TE: Can you talk about what specifically is your secret sauce at Avere? AS: Sure. Secret sauce number one is our internal file system that allows you to use all media types at once. Thats clearly the most important thing. And thats how you get our performance and capacity scaling separately. In addition, the second secret sauce is that when we decide something gets archived, instead of assuming that that archive data is in our box, we assume that its on someone elses mass storage system behind us. So the first secret sauce is the file system. The second secret sauce is that we allow other peoples filers to get pulled into our name space and to be a part of our file system the beauty is that we provide the performance envelope and any other vendors mass storage server provides the capacity. TE: And when you look at competition, are you competing really against the incumbents or do you also see small private companies trying to address the same problem? AS: You know, at this point, its really the incumbents. And the nice thing about that is the technologies that the incumbents are bringing to bear on this are not very novel theyre not very competitive. So if you buy an incumbent NAS filer and you ask for more performance, theyll sell you more spindles. And the crazy thing is theyll drive you to lower density drives because they need a lot more spindles in there to get the performance. So the incumbents typically talk about things like extra spindles, read caches which only help read workloads. The short answer is thats really our competition today. Were very competitive because we scale everything read, write and metadata. And because we allow our systems to plug into an existing NAS infrastructure, you dont actually have to leave an incumbents solution to leverage our system. You could just put us in front of it and drive the cost down of your deployed storage. TE: And do you typically replace a solution from these incumbents and who do you typically run up against? AS: So the great thing for our customers is that we dont require them to replace their incumbents solution. We typically run against the big players in the NAS space. So NetApp, EMC with Isilon, Sun and other servers. And typically what we show people is, if youre an Isilon user and you like Isilon, you can keep buying Isilon. Instead of buying their low-density, high-performance equipment, you buy their high-density, inexpensive equipment and you put us in front of it to get the performance back.

Page 47

March 14, 2012

Industry Report

TE: Interesting. Is there any specific segment or vertical that youve seen early traction? Its a very broad-based solution but are you seeing any particular industry or vertical take to your solution over others and why? AS: I think in terms of our market, its very much the markets where NAS is typically strong today. So, in the performance space it is media and entertainment, government, life science, financial institutions specifically financial modeling. In the WAN space, its really just large enterprise anyone that has multiple sites and is trying to get people to coordinate. In the cost space, its anyone thats ever bought a Fibre Channel drive because we showed them how to avoid it and get much higher density on the capacity side. Its very much the traditional NAS market segments where you see us play. TE: And can you share with us a couple of your key customer wins that are referenceable at this point in time? AS: One of our largest customers is Sony Imageworks. And Sony is using us both for performance and WAN deployments. Theyre coordinating multiple sites, but theyre also using us for media and entertainment typical apps like rendering. So thats two of our use cases. We have a seismic processing company, ION Geophysical, which put six of our nodes in front of dozens of very high-end NetApp filers and was able to achieve two and a half times the performance for their seismic processing. I can give you an example of a small customer. One of our recent wins is a school district up in Massachusetts called Belchertown. They were able to put us in a VDI (Virtual Desktop) environment a VMware environment where they had 500 users and boot storms were really crushing the performance of their infrastructure. They put two of our nodes in and they got back to the performance prior to virtualization. So theyre very, very excited. What you see is that we scale from very small customers up to very large ones, wherever there is NAS. And it really kind of attracts those three use cases. TE: And what is the go-to market model and who are your key go-to market partners? AS: Id say our go-to market model is very much that we leverage both our own direct sales force but also the storage resellers our storage channel partners. We really do sales through both direct and through our sales channel partners. TE: And can you talk about your technology partners? AS: I think probably the one where our relationship is growing and were doing more and more with is Intel. As a matter of fact, we use the Intel SSD devices. We started off with SLC (single layer cell) the X25-E parts and now we moved on to the 710 series the eMLC (enterprise multi-level cell) parts. That equipment has been absolutely phenomenal to work with. Of course, like everyone else, we use Intel CPUs and chipsets. So thats probably one of our key technology partners. In our go-to market strategy its really just storage-centric VARs (value-added resellers), VARs that are targeting specific applications where a core part of that application is storage. And were working our way into those solutions and theyre bringing their solutions to their customers. TE: And if you can go back in time and compare customers willingness to look at your solution, how does it compare to what you see today with what you saw a year ago? AS: First, I think, in general, customers use us for their competitive advantage. So, in other words, if they could do more processing if they can make a decision faster using our gear, that helps them. Customers will always buy to do that. As sites grow, when you have more and more users at more locations customers buy us to help coordinate that effort. The other thing that I see is even during down market times, when customers typically spend less because the economy is down, theyre always willing to find ways to save money. So that third use case we talked about cost savings if we can go in and all of a sudden the footprint in the data center gets smaller, its less upfront cost its less power, its less ongoing cost that they have to maintain. What we see is that segment is strong no matter what the market is doing. TE: And if you look ahead and look at the future what is the vision for the business, say, a few years from now? AS: If you think about those three use cases, thats all encompassing of the way people use NAS. I think our vision would be that we will have introduced a new NAS architecture where access requests to shared files and applications no longer go directly to your archive filer. Those requests will go to one of our appliances and thats where youll achieve very lowlatency and very high performance. And then our boxes will make sure that the data ends up on a classic archive filer that could either be local or somewhere remote, even off in the cloud. I think that if we bring our vision to market, its that. Its that we need to have a local appliance local storage processing to provide best-case performance independent of

Page 48

March 14, 2012

Industry Report

user or application location. And then your archive your bulk storage which has high latency necessarily because of the type of media in it, could either be local or remote. TE: And from a product roadmap perspective, can you share any new products offerings that youre working on that our investors should know of? AS: Sure. I think the current state of Averes product is that remember the way I described our product way back in the beginning which is you have users on one side and you have an archive device on the other, and you put us in the middle and we provide offload. Well what I love about where were positioned right now is we dont have to focus on any of those archive functions. We focus totally on their offload. Our offload numbers are very impressive. Today, a typical offload that we provide is approximately 50:1. So that means 49 times out of 50 we service their request locally in very low latency media. And only one out of 50 is the user exposed to the latency of the mass server behind us. And what we have all of our engineers working on today is that 50:1 offload - we want to better. So I think what youre going to see is that we are targeting to increase that offload by 10x or more. And as we can push that offload higher, the optimizations possible are greater. So you can imagine you can have even denser dumber storage for your bulk archive. Or you can extend the envelope of your enterprise and push it further and further around the globe and users will never see that offload. Our roadmap is all about how we could hide the mass bulk latency from the user and 50:1 is where were at today. And I think youre going to see us push the envelope and get that much, much higher going forward. TE: I know that you are a private company. But can you give us a sense of your size and historical growth rate? AS: As you said, were a private company. We dont typically share information and where we are in the market. But what I can definitely say is this we have hundreds of units out in the field, dozens and dozens of customers and theyre doing really well. And of course where we are as a company, weve been in market with the product for two years. Its all about footprint now. Its all about expanding the footprint and turning that hundreds into thousands. TE: And when you look out to the next few years, what do you think is your biggest challenge? AS: I would say, potentially, the biggest struggle is probably what any new enterprise vendor has is that our larger, Fortune 500, accounts are very conservative. These customers want to do a lot of testing. Theyre very thorough. They validate all of our claims. And so I think what you see today is the majority of our sales in those types of accounts involve a proof of concept that goes out first. We send evaluation equipment out, we prove it and then they buy. I think today probably thats our biggest challenge is to be able to convert customers using less POCs (proof of concept) going forward and to just make them very clear, clean transactional sales. I think thats probably our big challenge right now in our, go-to market model and what were working on improving that. TE: This has been an excellent overview of Avere. Is there anything else do you want to share with us to wrap up this discussion? AS: As Im thinking about it, if you really want to point people at a proof point, one of the very first things I talked about is our SPECsfs posting (http://www.spec.org). Let me describe what that is. SPECsfs is an industry standard benchmark all of the NAS players in storage use it. And what you do is you download this benchmark, it generates a typical enterprise load and you point it at your equipment and you see how high performance you can get. The top three leaders in Spec right now, Avere is number one, NetApp is number two and EMC with Isilon is number three. We are at 1.6 million IOPS. Then after that NetApp is at 1.5 million IOPS and Isilon is at 1.1 million IOPS. But not only are we at the highest performance but the telling part is that we pulled off 1.6 million IOPS using only two and a half racks. NetApp needed 12 racks worth of equipment to do 1.5m IOPS and Isilon needed seven racks of equipment to do 1.1m. Avere is the highest performance solution on a typical Enterprise workload with a fraction of the gear. Thats really what our customers are seeing today. TE: Interesting. Thats wonderful Ron. I appreciate your taking the time to talk with us.

Page 49

March 14, 2012

Industry Report

Kevin Brown, CEO of Coraid Coraid Inc. designs and manufactures Ethernet based SANs. It offers EtherDrive Ethernet SAN storage arrays; SAN management products, such as EtherDrive ESM1000 SAN Manager; EtherFlash, a solution that provides access to flash drives without the bottlenecks and limitations of legacy storage and Fibre Channel networks. Mr. Brown is an accomplished entrepreneur and executive, with experience in the networking, storage, security, and virtualization sectors. Most recently, he served as President and CEO of Kidaro, a desktop virtualization software vendor, where he led the team to a leadership position in this emerging segment. Kidaro was acquired by Microsoft in May 2008 and incorporated as a key element of Windows virtualization. Prior to joining Kidaro, he served as a vice president on the original executive team of storage security vendor Decru, where he led worldwide marketing, business development, and product management, and stayed on as a vice president after Decrus acquisition by NetApp. Here are excerpts of his interview with ThinkEquity LLC: Rajesh Ghai (ThinkEquity): Can you explain the key problem you are trying to solve in the data center at Coraid? Kevin Brown (Coraid): Coraid is developing a new class of storage solutions based around Ethernet and scale-out architecture. There are some pretty radical changes happening in the data center today in terms of workloads where people are moving to virtualization, with larger data set workloads and eventually, toward a very flexible cloud architecture. Its simply impossible to reach this endpoint using existing legacy storage technologies like Fibre Channel networking and big-iron storage arrays. These technologies evolved out of the mainframe era, and lack the flexibility, simplicity, automation and price-performance to keep up with big data and cloud workloads. The fundamental economics of silicon are always driven by volume, and the volumes of x86 processors, commodity drives, and 10 Gb Ethernet are all creating a compelling opportunity to redesign the datacenter. By way of example, the street price of 10 Gb Ethernet has dropped below $500 per port, providing an amazing amount of connectivity at low cost. Our founder Brantley Coile got started in the Linux market, and figured out a way to use massively parallel Ethernet to achieve faster performance than Fibre Channel storage, at one-fifth the cost. More importantly, this design eliminated multiple layers of complexity and cost, and provided a perfect fit with the elastic virtualization and cloud models of the modern data center. TE: Thats great. And why cant existing solutions from the incumbent storage vendors address the problem that youre trying to solve? C: Its really a classic innovators dilemma. Storage has grown into a $30 billion market, built around the complexity of the Fibre Channel ecosystem. Companies like EMC, IBM, NetApp and HP dont want to see that market commoditized or disrupted, but the pressure has been growing. This is analogous to a decade ago when Sun, SGI, and DEC were made obsolete by the rapidly improving economics of x86 processors and RAM, coupled with Linux and virtualization. The leading storage vendors have broad product portfolios that are segmented by price-performance and use case. The margin structures of their business wont allow them to adopt a simplified solution set, and the cost and time to migrate their existing functionality to a new architecture is significant. TE: Great. And can you help us define your target market and give us a sense of what the size of this whole market is? C: Well, this is a very interesting one. Coraid was started back in the Linux market back in 2004 and got built into the Linux kernel in 2005. We grew up in stealth in that market, and went from zero to 1,000 customers in the HPC (high performance computing), university, research and hosting markets. In the last couple of years, weve filled in a lot of the missing pieces in terms of functionality for the enterprise and we are now playing across a wide range of markets, including enterprise virtualization, cloud, video, and healthcare. Today, our solution is being configured everywhere from SMB customers all the way up to big scale-out clusters of multiple petabytes for large cloud, enterprise and government customers.

Page 50

March 14, 2012

Industry Report

Were actually addressing a very substantial amount of the total available market in the storage industry. This is exactly what happened with servers -- it used to be that youd have small, medium, large and extra large servers and they would only address a certain segment. Today, you have x86 servers that get racked and stacked as commodity building blocks. We think storage is going to end in the same place. TE: And can you talk about the key technology trends that you are leveraging to build your solution to take to market? C: There are a couple of layers of pretty substantial innovation. One is at the networking layer. If you look at traditional SAN storage, its become a real problem for a lot of customers. It takes up to 40-50% of the budget and has got a lot of complexity built into the protocols. Years ago, corporations ran a few large servers, and maintained an eight-port SAN (storage area network). Today, racks of x86 servers fill the data center, and youve got an 800 or an 8,000-port SAN. The level of complexity of configuring has gotten pretty overwhelming. The way that Fibre Channel and iSCSI (internet small computer system interface) storage protocols work is that theyre connection-based, so not only do you have to hook up the physical connections, you also have to configure the logical connections between every single server, every single switch, every single HBA (host bus adapter), and every single storage array. The complexity and inflexibility of legacy storage has become a huge challenge. Coraid has pioneered a connectionless SAN that matches the elasticity of todays virtualized server environments. Ethernet SAN is built on bare metal layer-2 Ethernet and as a result, we can eliminate all of that complexity and essentially make the entire SAN look like an internal drive to each server. This enables our customers to save not only on capital expenditures, but also the high operational expense of configuring and troubleshooting those systems. TE: Thats wonderful. And what is your secret sauce at Coraid? C: Well, weve got two layers. One is the networking layer. We developed and open-sourced a connectionless storage networking protocol called ATA over Ethernet (AoE), which has been native in the Linux kernel since 2005. It is essentially an un-protocol. Theres no interoperability needed between five or six different vendors like a traditional SAN. Its basically internal to our platform. Our second layer of innovation is at the storage operating system layer. Our CorOS is a distributed operating system, which is critical for enabling scale-out storage that can scale to petabytes without performance bottlenecks. Other storage vendors have struggled to retrofit scale-out design into their platforms, and its taken seven or eight years in some cases. Its very difficult to redesign your DNA to work in this distributed scale-out environment and thats something that weve had from the beginning. The combination of that connectionless, flexible network layer with a powerful scale-out operating system gives our customers the ability to start small and expand a system to handle large data growth, without disrupting any of the existing equipment thats there. The idea is to grow as you go; you never have to forklift upgrade, you never have to redesign your network, you just keep growing. TE: This sounds very compelling and I think its also very unique. Do you see any one of your nearest competitors trying to bring the same similar solution to market? Either public or private? C: Weve definitely seen some of the SAN vendors talk about scale-out, but really not deliver it. Even if they developed a scale-out operating system for their arrays, theyd still be running on the same old Fibre Channel or iSCSI protocols, which would block many of the benefits. On the NAS side, weve seen folks like Isilon and NetApp have been working on some technologies to give more flexibility, but the price performance of those solutions is still lagging behind what were doing. So we definitely see people moving in the same direction, but not with something that really works in a high-performance block environment, thats where weve got a big lead. TE: And can you share with us a few of your customer wins and how you went about winning those? C: Sure. We recently announced Sony Music as a customer. Theyre building a large music library and moving to offer a video cloud. They had previously been using one of those scale-out NAS vendors; when they looked at the overall longterm cost of operation, the architectural benefits of Ethernet SAN were compelling.

Page 51

March 14, 2012

Industry Report

The U.S. Department of Defense is another large customer, with petabyte-scale deployments in multiple groups. In one deployment, he last petabyte they bought from an incumbent storage vendor was about $11 million. We came in and did a buy a petabyte, get a petabyte free deal for less than $1 million, and they were able to reduce their operating expense by 90%. We did a number of these types of deals. Shutterstock is another top customer. They operate one of the largest online stock photography businesses, and we power a petabyte-scale system that is distributed over multiple data centers. TE: Good. And can you talk about your key technology and go-to-market partners and how you expect them to help you with this going forward? C: Sure. So were working closely with a number of partners. In the networking space, were working closely with Avaya and Arista and some of the next-gen 10gig players. We work closely with VMware and have a number of our executives that are alumni. Weve got a lot of projects going with the back-up vendors like CommVault and Veeam. We also work a lot with the file system vendors that can ride on top of our scale-out SAN on the back-end. People are shifting their thinking from thinking about just one layer of the IT system to really moving toward a cloud stack architecture. For example, were working a lot with the VMware channel -- theyve got a tremendous challenge with the cost of storage blocking progress on a lot of their projects. TE: Your solution is obviously pretty unique and I would say differentiated. What is your sense of that at this point in time? Are customers willing to look at a solution thats a little different and a little revolutionary compared to existing solutions from incumbent vendors? C: Well, Id say so. Weve grown our sales 7x in the last two years and we just passed 1,500 customers, including many multi-petabyte customers; our win rates versus some of the traditional vendors are quite high for many solutions, in the 60-70% range. People initially look at our value proposition around price performance and simplicity, but over time the largest value is the elasticity and flexibility of our architecture which matches what they are doing at the computing layer. In addition to the technology value proposition, the fact that weve raised $85 million in funding in the last two years, and weve got all of these use cases deployed in the field, puts us in a great position to get them to do something different. TE: Thats great. And can you talk about your vision for the business say a few years from now? C: Coraid is delivering a solution that combines the performance and reliability of enterprise storage, with the flexibility and cost structure of cloud systems like those built by Google and Amazon. These players think of computing at a different scale, where of the outside of the building is the computer and everything inside is just components. We acquired a company in October called Yunteq, which is orchestrating at the storage, networking and security layers to tie together all of those physical components into a system and to automate that in a very policy-driven way. This is exactly what the cloud guys have built, except they had to assemble 1,000 PhDs working for a decade to invent their own file systems and orchestration systems. We are delivering a commercially-supported version of that architecture, with enterprise class performance and data protection capabilities to handle mission-critical workloads. Its really cloud scale meets enterprise storage, and the purchase of Yunteq gives us a policy engine to wrap around that and to deliver a lot more automation. The consumerization of IT was a hot topic five years ago, but that trend is just now making it down into the infrastructure; end users want a system like Amazon where they can click a couple of times and have their whole infrastructure set up. Thats exactly where people want to get to and thats something that were in a very unique position to deliver on relative to the more antiquated Fibre Channel vendors. TE: Great. And when you look out over the next couple of years, what do you see as your biggest potential challenge for your company? C: Our biggest challenge is continuing to grow and absorb all of the demand that is out there for storage. We need to go build the teams and launch the territories and continue to scale up the company. And were very thoughtful about keeping a very high quality bar as were building the team. Its never easy to find the A+ players but thats how were looking to build the company.

Page 52

March 14, 2012

Industry Report

TE: This has been a great overview of Coraid, Kevin. Just to wrap up, is there anything you would like investors to know? C: Theres a major wave of architectural change sweeping all the way through the data center, starting at the computing layer. A decade ago many data centers were running big iron servers, and thats been replaced by scale out, x86, virtualized architectures. That wave of change is now sweeping down through the network layer. If you look at the street price of 10 Gb Ethernet ports dropping to $500, heading to $400, you can understand why some of the traditional networking players and certainly the Fibre Channel storage vendors are coming under a lot of pressure. The whole specialized industry of data center networking area is being commoditized today. Coraid is able to not only to leverage that lower cost per port, but also to use Ethernet in a massively parallel way because of our connectionless storage architecture. Finally, the same wave is going to break over the storage layer. Large scale-up storage arrays are going to be replaced with scale-out building blocks, built on commodity hardware. People are realizing that with the rapid data growth thats continuing, they just cant afford to buy a big fixed storage frame, fill it up and then forklift every 12 or 18 months as their data grows. Consequently, they also dont want to buy three years worth of capacity up front because it will sit empty most of the time given these growth rates. When these waves of change are complete, the enterprise will look a lot more like the cloud, and organizations will enjoy a level of agility and computing scale that will enable some major business benefits.

Page 53

March 14, 2012

Industry Report

Jean-Luc Chatelain, Executive Vice President, Strategy and Technology of DataDirect Networks DataDirect Networks, Inc. offers a portfolio of storage array, file system, and object storage appliances for big data applications for data-intensive environments, including high performance computing, life science research, Web and cloud content, professional media, homeland security, and intelligence environments. Jean-Luc Chatelain joined DDN in February 2011 with several decades of experience as a technology industry leader. Most recently, Chatelain was as a distinguished Hewlett-Packard Fellow and Chief Technology Officer responsible for successfully leading HPs Information Optimization strategy, steering the entry of HP into the ILM space. Here are excerpts of his interview with ThinkEquity LLC: Rajesh Ghai (TE): Can you explain the key problem you are trying to solve in the data center at DDN? Jean-Luc Chatelain (DDN): At DDN what we help our customers do is manage the vast amount of unstructured data that is coming at them and enable them to store, to distribute, and to process that information at extremely high scale. We focus on the high performance computing markets - the life sciences, the oil and gas, the intelligence community, the cloud, and media and entertainment verticals. What those four markets have in common is that they have been dealing with what is known now as Big Data, for some time now. As you can imagine an HPC (high performance computing) center manipulates terabytes of information. A movie like Lord of the Rings is terabytes and terabytes of information. The intelligence community has a massive amount of infrastructure data coming from various centers. And then in the cloud, we serve Big Data because of the fact that data velocity, ubiquity and scale are present at unprecedented levels. Our unique solutions allow our customer to be able to capture, store, and distribute, and process this information at very, very high scale and very high performance while maintaining a very low footprint in the data center where real estate is very expensive. TE: Why do you think existing solutions from the larger, more established vendors cannot address this problem? DDN: Because the traditional solutions, what you call the more established vendors, have been essentially IT-centric. They are excellent solutions but they are focused mostly on the transactional aspects. And they deal with structured and unstructured data that is very small in scale. We are, in our space, in a world where we count pieces of information in billions and capacity in petabytes. Its not that the competitive solution or more established market players have bad solutions that have been badly designed. Thats not the case. Theyve just been designed for a completely different workload. TE: My next question is something that you have probably touched on in your first answer. But are any other key market trends driving demand for your solutions? DDN: Yes, there is an amplification of some market trends. It is not a surprise that we are in the middle of an information explosion. That started in the mid 90s. This is mostly generated as a byproduct of the knowledge worker such as e-mails, and documents et cetera. But recently weve added new dimensions to that explosion. In fact, we are facing a new information explosion, mostly driven by video, by social media; but also by machine-generated data. Weve enabled machines to be mobile. Weve given the machine a voice. And they have a lot to say. They're very verbose. And that translates into a tsunami of information which demands more storage and this storage capacity demands more scalability in the number of information objects that can be managed. It also demands more bandwidth because its not stopping. TE: Can you give us a size of your addressable market that you are addressing with your solutions right now? DDN: Its very elastic. So depending on who you talk to in the cloud market, it is quite elastic in size. But I would say, conservatively, our market is $9-10 billion by 2014. It is a very conservative estimate of the HPC market - media entertainment, video surveillance, the federal market; and some aspects of the cloud. IDC estimates go as big as $25 billion in 2014 if you take the cloud as being more encompassing. TE: You talked about key market trends that are in influencing demand for your solutions. What are the key technologies that are enabling you to come out with a solution that is truly differentiated and unique? DDN: Number one, we base all of our solutions on an open system technology. So we use componentry that is used in the world of very high performance computing. In the industry, the key enabling technology, of course, is disk technology,

Page 54

March 14, 2012

Industry Report

which keeps on either augmenting in capacity, or given new performance metrics - such as the SSD (solid state drive) technology; which, you know, its not specifically disk but I would say is a storage technology. We see also a renewed interest for very high performance networks such as 10 Gigabit, specifically InfiniBand, which is getting a rebirth. With these open systems technologies, we go to great lengths that were maximizing their utilization of these components to exploit their capabilities and therefore the TCO to our customers by minimizing requirements. A very important trend is the need for in-storage processing. This was really not a key technology yesterday, but is a key requirement from our customers driven by the volume of data they are dealing with going forward. We are in an era now where the cost and time associated with shipping data to processing infrastructure is greater than the value extracted from that data. And we are really witnessing a transformation from data shipping to function shipping, where the customer wants to extract the value as soon as possible without incurring the cost of data shipping. Therefore, they want to virtualize their data-intensive applications inside the storage. We call this In-Storage Processing. This technology has the advantage of unleashing massive amount of compute directly where the data lives, where the data is captured without any latency or bandwidth limitations involved with the network. Other key technologies that are in our space include the emergence of object file systems that allow you to essentially break the scalability barrier of very large scale file systems. Traditional file systems do not have the capabilities to count in billions as I was mentioning a few minutes ago. And also, have a very large overhead, especially with very small pieces of data. Such as what is generated with machine-generated data. Therefore, a new paradigm is needed in how information is persisted in the file system and that is the object file system, which is an important technology that we are embracing. In fact, weve brought some interesting products to the market and we will see within probably five to ten years, that technology being the sole architecture in hyper-scale systems. TE: Your solution is pretty differentiated and quite unique. But in terms of competition, who do you see as your nearest competitors? And are you aware of any private companies or start-ups that are working on the solution? DDN: Who our competitors are is very clear. We compete with EMC and NetApp. There are also private companies that are attempting to do what we do, of course. This is a very dynamic space. So, there are a lot of people that want a piece of the pie. Based on our 14 years of experience in this market and having mastered the highest levels of scalability, we believe we have a very strong advantage. We have and will continue to deliver the worlds fastest scale-out and scale-up technology. No one has brought in-storage processing at the scale at which we can do, to market in a successful manner we are now, in fact we are in our second generation of this technology. And back to our two big competitors, EMC and NetApp, these are vendors that for a long time have ignored, to a certain extent, the market into which we play. But now, have realized that it is there and there to stay, and growing. I guess the best testimony to our market-space is the acquisition of Engenio, a former LSI subsidiary, by NetApp. I believe in their public statement, when somebody asked them why they bought Engenio, they said they wanted to go after DDN. One could argue that they didnt buy the right arsenal to fight this battle, but its a great backhanded acknowledgement of our position in the market. TE: And can you discuss a couple of your key customer wins and what helped you win in those accounts vis-avis your competition? DDN: Well, so, we have to go into specific markets because our customers are different. But we basically have more storage deployments across the top 500 HPC sites than any other company. As you know, we are the largest privatelyheld storage company. Key wins in the space include twenty-nine out of the top 50 HPC systems and over a hundred out of the top 500 according to top500.org. Key accounts in that space would be Oak Ridge National Labs, for example, the CEA which is the French Atomic Energy Research Center, the U.S. Department of Energy along with Taiwans National Center of High-Performance Computing. Those are some of the examples in the world of HPC. There are others in the life sciences. It could be Wellcome Sanger in the U.K. for example. In media and entertainment, we play in two spaces in the entertainment post-production, which is the making of movies and then broadcast, which is distribution to the air. A name you would recognize in that space would be Sony on the post production side. A client is also Deluxe, a name you are probably familiar with as at the end of most

Page 55

March 14, 2012

Industry Report

movies it says color by Deluxe. And then on the broadcast side, we have Fox Speed Channel, Canal+, and SkyTV in the UK. In the intelligence community and federal space, I would say you take any combination of three letters and you will get an agency that is probably a customer of ours but the U.S. Department of Defense is our main customer. We do a lot of work in that space. Finally and in the cloud, where we essentially are a supplier to cloud service providers, names you would recognize are Xbox LIVE, where we are the content processing infrastructure for Microsoft for their Xbox Cloud. Shutterfly also as a customer, it is a photo storage site not unlike Flickr. And then a company you are probably very familiar with would beYousendit.com which uses us. Those are examples of our customers. These organizations all chose DDN, and most are repeat customers, due to our engineered efficiency, broad and flexible product portfolio, simple scalability and our ability to understand the requirements of their very market-specific applications more so than any other company. TE: Switching to partners, who are your key technology and go-to market partners? And how exactly do they help you in your business? DDN: We are using open-standard type technology so key partners are the disk manufacturers. And from a technology point of view Intel is a good example of a partner as we use their technology on some of our implementation designs. When you come to InfiniBand there are different players. But partners include Mellanox and others. On the go-to-market, 80% of our revenue goes through the channel. We have very large, global partners, such as IBM, HP, Dell, SGI and Fujitsu. And then we have a growing network of other partners that tend to be specialized either on a by vertical basis or by geographic coverage. So we have a leveraged model powered by a suite of partners that represent the company globally. They're selling in that space. They speak the customers language. Unlike classic OEMs or broad market channel driven companies, DDN prides itself on knowing every single one of our customers. Its very important for us to have a one-on-one relationship with the end customer, because the only way you can build the right product is to be talking to a customer directly, without a proxy. TE: Turning to the future. What is the vision for the business say, five years from now? DDN: We have a clear vision statement It is timeless, it is what guides us. Our vision is to enable organizations to maximize the value of all information everywhere. We understand that the future of our customers data strategy is to derive insight and results from information. Our objective is to be the player that can offer a collection of information analytics products and solutions to each of our customers. DDN is in the process of a transformation of what I will call going from being an extreme storage company into an information solutions and analytics company powered by extreme storage. We have the DNA of understanding what it is to capture, to store, to distribute massive amount of data at the highest scale. We have the technology to in-storage processing and accelerate understanding - a big paradigm shift in this space. And we want to leverage this infrastructure and use it with smart software that will help customers extract value out of their information - because that is the reason why they keep information. If they did not believe there was a value to the information they are storing, they would throw it away. DDN will evolve to be the engine of their growth - able to help our customers transform that noise into a signal that can significantly change their business. TE: And what do you see as the biggest potential challenge for your company in the near future? DDN: Growth is our biggest challenge - in the sense that we have to have the human capital to be able to fuel that growth. We need to continue executing to our strategic staffing plan. Right now, were working on scaling the company to become a $1 billion+ systems and services providerTo get there, the biggest challenge is human capital. Are there technical challenges? I think there are, but I think we also have the smart people with a lot of experience in this space. TE: Can you give us a sense of your size and historical growth? You talked about the $1 billion goal. But what is your size at this point of time? DDN: We are 400+ people strong. 2011 will be greater than $200 million in annual revenue. We continue to experience double-digit growth and we have a commensurate increase in profits. We are a very healthy company with no debt - it all just comes down to the matter of execution. TE: Jean-Luc, thanks for your time. I wish you all the best.

Page 56

March 14, 2012

Industry Report

Gareth Taube, VP of Marketing at Kaminario Technologies Kaminario, Inc. provides high performance solid state storage devices. It offers the K2 storage appliance, a performance data appliance to support high IOPS, low latency applications without compromising on high availability. The companys solutions are applied in databases, such as Oracle and MS SQL Server; data mining and warehousing; business intelligence; operational business intelligence; customer relationship management; and Web. Gareth Taube is responsible for the company's marketing strategy and execution. Prior to joining Kaminario, Mr. Taube ran Market Recognition; a marketing consulting firm specializing in developing market strategies for VC funded early stage companies. Previously, he was an SVP at Emulex, where he ran the IP Storage Networking Group. Here are excerpts of his interview with ThinkEquity LLC: Rajesh Ghai (ThinkEquity): Can you explain the key problem you are trying to solve in the data center? Gareth Taube (Kaminario Technologies): The access to data at faster and faster speeds is driving business today. Most data center technologies have increased in capabilities over the past few years. One technology, spinning disks has not kept pace. According to industry analysts, many of todays application performance problems are caused by long I/O (input/output) wait times. These wait times are driven by mechanical disks spinning and their arms moving. Speed up storage and you can speed up your business. TE: Why cant existing solutions from the larger more established vendors address this problem? KT: Todays major storage vendors built their solutions for spinning hard disk drives. They architected their arrays for the issues HDDs (hard disk drives) have. For example, they are controller-based with large amounts of cache for holding data while waiting for reads and writes to complete. Most storage architectures by EMC, IBM, HDS, et cetera are at least 20 years old. These vendors are attempting to modernize their architectures but this takes time and they have a current revenue stream to worry about. Joe Tucci at EMC recently stated that EMC is behind the curve in new SSD technology. TE: What are additional key market trends that you see influencing demand in your target market? KT: Data quantity is exploding. Businesses are becoming global and there is a demand for more and faster answers to key business questions every day. In addition, there is a major revolution going on in data center storage. The cost of SSD (solid state drive) media has been dropping rapidly in the last few years while its reliability has increased dramatically. With the cost of SSD media declining, it is now possible to use this media in enterprise storage arrays. This price decrease and improvement in reliability has brought SSDs to the forefront of storage for the enterprise. TE: Can you give some sense about the potential size of your addressable market? KT: By both Gartner and IDC estimates, the Enterprise data center storage market was around $18B in 2011. That market needs speed more and more. It also needs reliability, availability and a price point that wont blow budgets out of the water. New SSDs have the potential of meeting that need and capturing an ever-increasing part of that enterprise storage market. TE: Can you provide some more detail as to how your solution addresses the problem? KT: The Kaminario team has deep storage expertise and understands the capabilities and limits of the new SSD storage media. They know that current storage architectures were not designed for SSD media. To solve this problem they have created a new type of storage architecture specifically designed for SSD media. This architecture is called SPEAR (Scale-out Performance Architecture). This delivers blazingly fast speeds at ultra-low latencies, self-healing for built-in data protection and high availability, easy to install and use, and is lower cost than existing SAN (storage area network) storage. TE: What specifically is your secret sauce? KT: SPEAR is the Kaminario secret sauce. Kaminario has designed SPEAR with the enterprise buyer in mind. It is blazingly fast with the lowest latency and highest throughput on the market. Its safe as it is equipped with automated selfhealing high availability to protect data. It includes internal mirroring, snaps for backup, replication for disaster recovery, a special RAID 10HD across nodes and data types, and full N+1 redundancy. Its easy to install and use. Kaminario K2 installs in just a few hours and, as a standard SAN, it requires no changes to existing infrastructure or applications. Finally,

Page 57

March 14, 2012

Industry Report

its at a low cost to the customer. Kaminario hasscale-out architecture. You can cost effectively scale capacity, performance, or both incrementally. It offers more performance with less footprint and power than existing SAN solutions. TE: Who do you see as your nearest competitors and then, more broadly, how do you see competition evolving? Do you typically replace a solution from one of your competitors? How intense is the competition right now? KT: Kaminario was the first in the market with production quality enterprise SSD storage arrays. We have been shipping product for over 18 months. In that time we have seen a number of all-Flash vendors come into the market but none have the breadth of features that Kaminario has. TE: And how do you sustain your competitive advantage in the market? KT: Kaminario has a long roadmap of additional features and performance enhancements that will be added to the product. For example just this past September Kaminario introduced the industrys first all-SSD SAN storage with media freedom of choice. In one Kaminario array you can have all DRAM storage capacity for the worlds fastest writes, all-Flash capacity for fast reads and low cost, or the Kaminario Hybrid with a mixture of both. TE: Who is your target customer and is Kaminario targeting any specific segments or verticals? KT: Our customers are across all industries. However, most are financial services, media, or manufacturing companies running a major database, SAP, or SAS application. TE: Can you share with us a couple of your key customer wins and how those were won? th KT: Some of our recent wins include a major electric utility company, the 8 largest bank in the world, a major media company, and a major university. TE: What is your go-to-market model? KT: As an early stage company we are 80% direct in the US and we sell through resellers in Europe and the Middle East. Recently we signed a worldwide reseller agreement with Dell and a cooperative marketing agreement with Fusion-io. TE: Who are your key technology and go-to-market partners and how exactly do you expect them to help you in your business? KT: Dell is a global partner as we build our arrays out of Dell components. Fusion-io is also a major partner as we use their PCIe (PCI Express) cards for our flash systems TE: At this point of time, how are demand trends and customer willingness to spend on a new solution such as yours? KT: The market for SSD storage is continuing to accelerate. Customers are starting to understand the need to add SSD storage to their datacenters; the only question is when and at what cost. TE: What is your vision for the business, say five years from now? KT: Kaminario is in one of the hottest markets in years. We are focused on building our business in this rapidly growing market. TE: From a product roadmap perspective, what new product offerings or features and functionality that youre working on that our investors should be aware of? KT: We cannot disclose any unannounced features but later this year we will enhance our data protection and high availability capabilities and well as reduce our costs and footprint. TE: Can you give us a sense of your size and historical growth rate? What are your revenue projections for the future? KT: We were founded in 2008, shipped first product in 2010 and have been on an aggressive growth path ever since. Unfortunately, we cannot discuss specific sales or revenue figures at this time. TE: When you look out over the next couple years, what do you see as the biggest potential challenge for your company?

Page 58

March 14, 2012

Industry Report

KT: Probably handling the growth that we anticipate to come with our product and based upon overall demand within the market. TE: Is there anything else you would like our investors to know? KT: Kaminario is a leader in the latest storage revolution. Our software portfolio and our SPEAR architecture will make us a major force in this data center transition.

Page 59

March 14, 2012

Industry Report

John Spiers, CEO and Founder of NexGen Storage NexGen Storage, Inc. provides shared storage systems. Its enterprise class storage systems are used for server virtualization, virtual desktop, business critical applications, and analytics solutions. John Spiers is CEO and Founder of NexGen. Prior to founding NexGen Storage, John was founder and CTO at LeftHand Networks. LeftHand was sold to HP in 2008. Prior to founding LeftHand, John was director of software engineering at Maxtor Corporations Network Systems Group. Here are excerpts from his interview with ThinkEquity LLC: Rajesh Ghai (ThinkEquity): Can you explain the key problem you are trying to solve in the datacenter at NexGen? John Spiers (NexGen Storage): One of the main visions for the company was to focus in on some of the problems that we did not solve when we were at LeftHand Networks and some of the problems that started to emerge from the growth in storage and server virtualization. And one of those main problem areas was around virtualization and control of performance in a virtual server environment. When you consolidate your server and storage resources into a shared storage system, that shared storage system is being shared by multiple applications simultaneously. Nobody out there can control the performance being allocated off that shared storage system back to a given virtual machine or application at any given time. For example, if you have a spike in your marketing file share that may impact your order entry database or other critical applications. What customers were doing was they would throw a lot of 15K (RPM) disk and overprovisioned performance into their storage array, so that when those spikes did hit the system it didnt overwhelm the system and kill all the other apps that were trying to use the same system at the same time. We are really focusing on solving that remaining problem in shared storage. Were also taking advantage of new technology that is in the marketplace. There is a lot of buzz obviously around solid state. Today, most of the storage solutions are what we call solid state bolt-on solutions. So, theyre taking solid-state technology and bolting it onto legacy disk drive architecture. And there are bottlenecks up and down the I/O stack starting with the drive interface that connects to the storage controller which is also a bottleneck. And even the software stack and RAID algorithms and the list goes on and on. One the things that Fusion-io identified early on was that solid state is very fast, close to as fast as memory, and it doesnt belong behind a drive controller. It belongs right on the compute bus. We believe in that also. Weve designed our shared storage system to deploy solid state on the compute bus. Our SAN (storage area network) is powered by Flash. We think of our solution as a shared Fusion-io solution where multiple servers can use the Flash simultaneously because it is a storage area network versus Fusions current deployment model which is Fusion inside a server and more of a direct attached storage model. We are delivering quality of service for a versatile server environment and solving that performance problem to determine the performance. We have phase data reduction which is our deduplication technology. It allows you to get up to 5x improvement in usable capacity through deduplication. Another problem we are solving is the storage efficiency problem. We can deliver much higher performance, at a lower footprint, and typically deliver a 90% operating expense savings to customers as far as footprint, power, dollar per gigabyte, dollar per IOPS (input/output operations per second). TE: I think you briefly touched upon this. But can you tell us why don't existing solutions from the larger, more established vendors address this problem? NS: Shared storage environments were designed from the ground up to consolidate the management of capacity and allow you to have a single resource in order to get higher utilization. None of those existing solutions, in terms of NAS and SAN offerings, were really designed to control performance. Because theyre not architected to do that, its very difficult to provide a bolt-on kind of management application to be proactive and provision performance like capacity. What weve done is re-architected that system with the number one goal, from a technology perspective, of controlling performance in

Page 60

March 14, 2012

Industry Report

a shared storage environment. Its a really hard thing to do and, because weve rearchitected that, that is how we are able to deliver this and existing solutions cannot. TE: I understand that you are targeting the primary storage market. Can you give us a sense of the size of the market that could be addressed by this solution? NS: Like you said, you start at the generic primary storage market. External storage systems are a $30 billion market. We target customers that are not SMB, they are not the small guys and they are not Fortune 500. Its a pure mid-market play. Even further, we look at the virtual infrastructure maturity model, like how far along are they in virtualizing their infrastructure. We then look for what we call advanced virtualization shops. Those are people that have virtualized once and have probably done it for their non-critical production or test deployments. Now they are looking to take the next step and increase their virtualization ratios and make it a strategic part of their ongoing management of the datacenter. If you size the storage market to support something to that effect, we think the market is between $9 billion and $15 billion. TE: You talked about Flash as an enabling technology. Are there other technologies that are enabling the solution today that you may not have had access to in the past? NS: There are a couple. One is the higher bandwidth that is available from a network perspective. 10-gig (10 Gigabit Ethernet) has been around for a long time. But we are finally starting to see it hit mainstream. And what a lot of folks are doing, and again these are the advanced virtualization folks, is that they are looking to consolidate. So they leverage the bandwidth that they can get on 10-gig, so that is one way to approach it. The other way that enables our system is the increase in bandwidth in the computing bus, specifically around PCI Express. The ability to move such massive amounts of information in such a short period of time and do it on a compute bus really is at the core of allowing us to do what we need to do in our system. Were also taking advantage of faster memory and faster processors. Our system has 48 cores of product capability in it. TE: Can you briefly tell us what exactly is your secret sauce compared to your competition? NS: Our secret sauce is the ability to dynamically tier data to deliver service level performance. A lot of the tiering solutions out there are what we call after-the-fact tiering. It is very difficult to manage your policies for something that happens at an instant in time and then hoping that that policy works going forward. When blocks get hot within a product such as Compellent, they setup a policy that tells the system that if blocks of data have been hot for a given period of time, you move them into a faster tier. When blocks of storage are cold for a certain amount of time, you move those to slower storage. The problem is that as a result of those block movements and policy changes may or may not deliver the service that you are expecting. For instance, what if you have an end-of-the-quarter report that kicks off for your billing? Chances are, and since it only happens once a week or once a month or once a quarter, that data is going to be on slow storage. So, now that application is going to suffer. There are always random events that occur that require higher performance and service levels, so the way we tier data is to do it dynamically based on a performance setting. If a customer says he wants X-performance out of the system, we dynamically tier the data on the back-end to deliver that performance level. And we allow that performance to spike if the overall system resources are available to let it spike. Another thing we also do is tier data based on an event. In that example I just gave, you know these queries are going to kick up on every Friday afternoon, you can actually schedule that tier in operation so that when those queries run, that data is in the fastest tier. The main advantage of dynamic data tiering based on quality of service is the ability to operate based on time. And then we have a phased reduction deduplication capability which is dedupe without a performance penalty. There are very few systems in the market today that can do that. Most people do dedupe post process. We do it on primary storage without a performance penalty. And our dedupe engine is tied into our quality of service engine; so that we are not burdening the system with dedupe activity while we are delivering a service level. That service level takes a priority over that operation. Our secret sauce differentiation is architecting the system from the ground-up, delivering quality of service, better storage efficiency for deduplication and very fast performance through a PCI Express- based back plane architecture. Not only do we scale up with PCI Express but we scale out cluster over PCI Express as well. PCI Express is a very high speed, low latency back plane that rivals InfiniBand and other solutions out there.

Page 61

March 14, 2012

Industry Report

TE: That sounds very compelling and very different. If you look at the storage landscape, do you see anyone that has a solution that is close to what you have? And I am not just talking about the established large players but also the startups that are beginning to emerge in this landscape. NS: From a large player perspective, they are probably the furthest behind us because theyve got so much momentum behind the existing technologies in place. They need to retool their architecture to get to where we are. We see lots of differentiation versus those guys. In terms of the new startups, you will see elements of what we are doing in different segments. For instance, there is a SAN provider that is putting Fusion in their system. They are called Kaminario. Their systems start at half a million dollars so theyre very much in the high tier and they target specific niches that only care about performance. Then youve got a lot of other guys that are doing inline de-duplication because they have all solid state systems and they can get around latency issues. Their dollar per gigabit equation is not nearly as good as ours. We see there are elements of differentiation that are cropping up in different market segments. But no one has put everything together and targeting the mid-market like we have. TE: Thats great. Who is your target customer? Is there a specific segment or vertical where you are seeing initial traction? NS: We look at it as more of a horizontal solution. You can call virtualization a vertical but you know that everyone is virtualizing - so it is hard to do that. Its very horizontal. And the next layer down for us is really the advanced virtualization market - the folks that have done it for two to three years and are taking the next step. Not necessarily the customers that are just starting out. TE: Unlike LeftHand, you are not really focused on the mid-market or any specific market size or particular customer segment? NS: No. It really has to do with the virtualization maturity. TE: Is your product shipping at this point in time, and if it is, can you share a couple of your key customer wins and how you went about winning those? th NS: Sure. The product is shipping. On November 8 our company came out of stealth mode along with general availability of our n5 storage system and weve got a couple of customers we are working through some agreements - that we can do joint marketing with. We have not been sharing names to this point but we will shortly. One example is a nationally known museum that had a SAN that was running the point of sales system for them along with some back-end business intelligence apps. What they had to do was tell their business users that they could not run reports during the day because it impacted the point of sales system. With NexGen, they can provision the performance for the point of sales system so they know that if they run another application, the point of sales system is still really fast. Now their business users can run reports during the day and not impact the customer experience via the point of sales systems. That is just one example. There is another example of a research institution that needed to consolidate their datacenter footprint by 50%. Not only did we help them do that, we took them from 11 racks down to three racks, because now they could virtualize rest of their applications and save that server footprint. But more so, we consolidated 42U of scale-out storage down into a 6U footprint with our solid state disk drive. They are now seeing 90% opex savings in terms of not having to do drive maintenance agreements and the power and cooling that goes along with that. TE: At Left Hand, you had a channel go to market model. Are you following a similar approach over here or is it going to be different, given that you are not focused on the mid-market here? NS: We are actually still going to take it through the channel. We think that the channel actually reaches up into the SME, or Small Medium Enterprise, and we can make hay there so were still going to use that go to market. TE: You mentioned Fusion-io. Are there any other key technology and go-to-market partners that you have that you can talk about and how exactly do you expect them to help you in your business? NS: Absolutely. From a Fusion standpoint, they are very excited, as are we, about the relationship. We had a press release where we announced the relationship of joint sales and marketing activities. They see us as an incremental market to what theyre going after. They go after a lot of Web 2.0 and very high-end kind of analytics environments. We can open up the mid-range SAN market for them and allow them a shared storage solution. What that enables now is that you can get Fusion-io in a shared storage system and you can do all of your virtualization and you could run vMotion. You could share Fusion with all of the application servers connected to that shared storage system. That is something that

Page 62

March 14, 2012

Industry Report

doesnt exist today at the mid-market price point. They are very excited about that and we see a lot of joint engagement in the field to find those new opportunities and bring a new market to them. TE: You have been out about a month now, what is your sense as far as customer willingness to look at a new solution such as yours? NS: Its actually been tremendous. What happens in a sale is we will go through the pitch and when they see the quality of service capabilities and ability to manage performance; you see a light bulb go off. Immediately the technologist, from a customer standpoint says, I cant do that with any other system and that is a big problem. They immediately are latched into it. In fact, a lot of times, we just end the presentation there. And because performance has been such a big problem, I think a lot of people have given up and just accepted the issues. But when we come in and say no, you can actually solve this and manage this now, it is something unique and different that they havent seen. They immediately want to try it because it is so different. We have seen a lot of acceptance, in fact, weve been self-throttling a little bit, because, were building up our field team. We want to make sure that we have all of the pieces in place so we can ramp and scale with customer demand. Its been tremendously successful thus far. TE: Looking into the future, whats the vision for the business five years from now? NS: We have a lot of stuff on the roadmap. We plan to support Fibre Channel and NFS on our system and well also scale out. Today we just scale up. We allow customers to increase performance by adding more Fusion and that increases capacity by adding ourselves from the back. What is nice about it is that you can scale one or the other independently. Then our PCI Express scale-out capability will be introduced next year which will allow us to expand, solid state and solid state bus attachments on a scale-out basis which will allow us to potentially move up market and deliver much of our performance. We will see some new partners as we do that. TE: As you look over the next couple of years, what do you see as your biggest potential challenge for the company? NS: I think there is a lot of noise in this space. Everybody is going out with solid state. Were going to see a massive transformation in the storage industry. The key is that it is still going to be much more expensive than rotating disk. Some people believe that just by deduping data on the solid state that they can be competitive in this space. But if you look at a very intelligent tiered system that also does dedupe, instead of going from $10 per gig to $3 per gig, were going from $3 per gig to $0.70 per gig. So it is a whole new dynamic in the marketplace if you are able to tier very efficiently and deliver a very high service level and still have disk in your system, youre going to outcompete the guys that are 100% solid state. So we do not see the market shifting to 100% solid state. Maybe its a high end and niche market but mainstream storage guys that need 100 terabytes of storage and doesnt want their idle data sitting on expensive solid state is going to be more attracted to a solution like ours. TE: Just to wrap up, is there anything else that you would like our investor clients to know? NS: A lot of us have experience with building up the scale-out iSCSI market so weve kind of gone through this once already and are very familiar with the ramp rates and the tactics used to hit the growth that companies like Compellent, EqualLogic and Left Hand did, so there is a lot of experience here along with that. I think when we look at NexGen, we think we have kind of experienced a lot of the mistakes already and know what it takes to be successful. We are extremely bullish that we can ramp and make this a much larger business, much more quickly than what occurred in that first round of the iSCSI scale-out that happened a couple of years ago. From that perspective, I think we can grow a lot faster into a larger company on a much shorter timescale. If you look back at the last 10 years, we were evangelizing iSCSI, evangelizing scale-out and it wasnt until Microsoft endorsed iSCSI that we really started to get market traction. And the good news now it is not a new technology, per se, to evangelize, but is more of solution to problems around virtualization that havent been solved and delivering a storage system that is much more efficient and delivers a better value for the customer. TE: Thank you for taking the time to talk with us.

Page 63

March 14, 2012

Industry Report

Suresh Vasudevan, CEO of Nimble Storage Nimble Storage offers a storage and backup systems solution that significantly reduces the cost and complexity of storing, accessing, and protecting data for enterprises. The company's CASL architecture uniquely combines flash and highcapacity disk to support the most demanding primary applications, while also reducing backup windows and restore times from hours to seconds-all for at least 60 % lower cost than existing solutions. Mr. Vasudevan joined Nimble Storage as CEO in 2011, and has served as a member of its board of directors since 2009. Most recently, he was CEO of Omneon (acquired by Harmonic), and previously served as a member of the executive team at NetApp, overseeing all product operations. During a decade-long career at NetApp, Suresh led the companys product strategy and product development and was a key architect of the steady expansion of NetApps product portfolio into new markets. Before joining NetApp, Suresh worked at the management consulting firm McKinsey & Co. in New Delhi, Mumbai, and Chicago as a senior engagement manager. Here are excerpts of his interview with ThinkEquity LLC: Rajesh Ghai (ThinkEquity): Can you explain the key problem you're trying to solve in the data center? Suresh Vasudevan (Nimble Storage): Nimble Storage as you know is a storage systems company. When we think about what we do differently you have to think of how data centers deploy storage. Storage in a data center context - part of it is comprised of a primary storage tier which is where the application houses its data. Then you have a place where you make a copy of that primary data and that is often a backup system - traditionally tape drives, increasingly disk-based backup systems. Back-up systems protect against loss of data in the primary storage system. And then, in case you have a disaster and to protect against disasters you often also make a copy of the data that lives off-site and that represents a disaster recovery solution. So thats the typical mental model for how storage is deployed in data centers. When you think about all the mainstream storage solutions today, there are at least a couple of significant issues that existing storage solutions dont tend to address really well. The first of those is what I think of is a tradeoff between performance optimization and capacity optimization. So let me explain that. When you think about deploying storage for applications, most storage vendors today start the conversation by saying, Are you solving a performance problem? or Are you looking to solve a capacity problem? Examples of performance problems are how many Exchange users can I consolidate before my storage system runs out of performance? How many virtual desktops can I support before I run out of performance and have to deploy a bigger or another storage system? Those are all instances where the storage system is sized based on performance needs of the applications. There are other environments where you are solving for capacity. For example, I am storing a large amount of documents and Im trying to optimize the cost per gigabyte of the storage I deploy. There you're concerned about whats the cost that the storage system presents for a given amount of capacity. Now, thats typically how CIOs and IT directors think about their storage needs and traditionally most storage systems either optimize for performance, which means that they are solving for a given amount of performance but they end up wasting capacity. So its efficient when it comes to the cost of delivering a certain amount of performance but theres a lot of excess capacity that you're not able to use. Or if you're solving for capacity, you do get the amount of the capacity you need, but its not good enough to run performance applications. So thats the first problem that we see with every storage system today. You either get performance efficiency or capacity efficiency but not both at the same time. The second big challenge that we see is that in this mode of deploying storage where you have primary tier, back-up tier and then a disaster recovery tier. The big challenge that customers face is that, in order to make a back-up copy, they have to copy all the data from their primary storage system to their back-up storage system and they often have to do this within a window of time often called a back-up window and moving along a large amount of data takes an enormous amount of time. Secondly, when they have an error and they need to recover data from their backup system, they again have to copy all the data back and that takes time. So recovery tends to take time as well. This is also costly because you have separate backup gear. Companies like Data Domain were able to innovate and move backups off of tape on to disk-based storage systems. So they solved the tape component of the backup problem. What they could not address was this constant need to move

Page 64

March 14, 2012

Industry Report

data from primary to backup and move it back. And thats really where theres a significant amount of pain. So the second challenge that we address is the problem of moving data and the inefficiencies associated with time to backup and time to restore. We do this by consolidating both the primary and secondary or backup data into a single system. TE: Yes it makes a lot of sense. The question that often comes up is why havent the legacy vendors, the more established incumbent vendors, looked at this problem? Or why dont their solutions address the problem right now? NS: So let me take both of these problems and say why do these problems exist? And Ill talk a little bit about how we address it but let me start by saying why have they not been addressed thus far. Performance versus capacity is simply an artifact of how the storage industry has evolved. For about 20 years the ultimate building block of every storage product line that exists today has been disk drives. So there is a lot of software intelligence built into a storage system but ultimately the place where you store your data traditionally for the last 20 years has been a disk drive. When you think about the characteristics of the disk drive, disk drives have over the last ten years gone in density from about 18 Gb to about 3 Tb. So theyve gone up by almost 150x in terms of density. And yet the speed has only gone up by a factor of two in the last ten years. If you translate that, what storage systems have had to do is, in order to deliver performance, theyve had to gang together a large number of disk drives, each of which serves some amount of I/O. Thats how they deliver performance. Essentially by having more and more disk drives, each one contributes to your I/O performance. But in doing that, you're adding more disk drives than you might need for the application, from a capacity standpoint. So theres a fundamental tradeoff in order to deliver performance, you need to add a larger number of spindles or disk drives. That creates excess capacity. If you need the performance you don't have a choice but to deliver that excessive capacity. Thats been one major reason why most storage companies have not been able to address this performance vs. capacity tradeoff. The second problem really is around the data protection problem that I mentioned earlier. When you're solving for backups, you're looking for something that has very low cost. And since primary storage systems have always been all about delivering high availability, higher reliability and higher performance, they are not a good medium to also store back-ups because this is too expensive for backup data to be stored. Thats why when Data Domain came around they created a completely separate disk-based system using low-cost SATA (serial advanced technology attachment) drives, using compression and deduplication to lower the cost of the storage system. In turn, they were able to move all the backups from tape to a disk-based system. But they had to live in a separate cost-optimized storage system. It could not live in the primary storage tier because it was too expensive for that. Thats why youve always had data protection as part of two separate tiers, the primary tier and a back-up tier. So if you really wanted to break that sort of gap between primary and backup tier, you would have to make your primary tier not just good at performance, youd have to also solve the cost aspect of primary storage. If you can make the cost of primary storage match the cost of backup storage, then theres no reason for the two tiers to remain separate. So these are interlinked problems. If you can sort of, simultaneously solve for performance and cost of capacity then you can go on to address collapsing both primary and backup data tiers as well. TE: Great. Are there any market trends that have caused this problem to come to the forefront that gives you confidence that the solution that you are creating right now will have an addressable market thats significant? NS: Ill say this - the single biggest market trend that has brought this problem in starker contrast for customers is server virtualization. In the past before server virtualization, you were often deploying storage in the context of a single application. With server virtualization there has been a change. First servers have become more powerful. Coupled with this is the ability to have a single server run multiple applications on different virtual machines on that server. This has meant that the I/O thats targeted at any storage system has become more than traditionally the storage system had to address - so theres more I/O demands in the storage system. And more importantly, because on the wire you're now mingling the I/O needs of a variety of different applications, the I/O pattern has become more random. You're blending together many applications which increases both the absolute I/O that you need to serve up from your storage systems perspective and the randomness of the I/O thats being demanded of the

Page 65

March 14, 2012

Industry Report

storage system. Now if you translate that, what it means is that storage systems have traditionally often been purchased by customers on the basis of dollar per usable gigabyte. Thats often been the driving reason or the driving basis by which storage systems have been purchased. Increasingly, thats not a good metric as customers are more and more indicating, Look I cant just live with certain amount of usable capacity, I also need a certain amount of IOPS (input/output operations per second) and dollar per IOPS has become an increasingly important metric as well. So in many ways, server virtualization has caused this performance versus capacity tradeoff to become critical for customers. The other problem that I mentioned is data protection and how to constantly live with backup Windows and restore times this has also been exacerbated by server virtualization. The whole notion of deploying a back-up agent and then using a back-up application which moves the data through the server that is hosting my primary application to a back-up device was okay when you had multiple applications all on their own dedicated servers. However, as you mixed many applications on to a single, large server, you've raised the CPU utilization of that server because you are very effectively able to use the same server for multiple apps by using server virtualization. What that has meant is that this whole agentbased backup approach where a backup app is loading agents and then copying data through the server on to a backup device is becoming much more difficult. You don't have the CPU to dedicate to a backup process. You don't have the backup Windows within which you complete your backups on time. So thats become a very big problem as well. So in many ways, server virtualization has really caused both problems to become much more prevalent among our customer base. TE: When you look at your addressable market, are you thinking of the market as including both primary and backup storage? NS: Thats a great question. The way we think about our market is we address primary storage and data protection uniquely with our approach - but the market opportunity for us is the entire iSCSI (internet small computer system interface) storage market. So thats the primary storage market and that, in itself, is roughly a $3 billion market. Associated with that is storage software. We tend to bundle all our software with our storage system into a single SKU. Having said that though, theres a unique market that can be measured for things like storage software that goes along with these storage systems and thats an incremental $1 billion. Without even counting the disk-based backup target market, were looking at a $4 billion market opportunity in storage hardware and storage software. You can then add to that some component of the disk-based backup market thats currently being sold to by companies such as Data Domain and ExaGrid. Suffice to say, we think were dealing with a market thats measured in $4-5 billion in size and growing well in double digit percentages. iSCSI is growing somewhere in the 20% range and is a fairly substantial market that were dealing with today. TE: Can you talk about the key technologies that are enabling you to bring your solution to the market? NS: Let me start by suggesting that there are a couple of disruptive trends that we are leveraging that weve been built our software stack around. Let me start with a couple of disruptive technology trends themselves. The first one is of course, flash and solid state drives. In many ways I think they are singlehandedly the most radical disruption weve seen over the last 15-20 years in our industry. And call into question almost everything that has gone into traditional storage architecture because traditionally storage has been architected around disk drives and how to optimize disk drives as the underlying media. So that is the first disruption. There are a couple of other disruptions as well that we have leveraged quite heavily and they may look commonplace today but when you think about when they came to be, you'll realize that even the most modern storage systems within companies that started four or five years ago were not optimized around those. The first trend is availability of multi-core CPUs. Today we think of multi-core CPUs as fairly common but yet if you go back and look at the architecture for Dell, EqualLogic, Compellent or even go further back and look at NetApp or EMC, they were not necessarily designed with lots and lots of cores as a given. So thats one trend. The second trend that we leveraged were low-cost drives, low-cost near-line drives, SATA or increasingly near line SAS (serial attached SCSI). Again, something that became very prevalent in 2004 in enterprise storage arrays, but prior to that, it was only used in desktop drives. So those are three component trends that we leveraged and how we built our system software ground up. Having said that, what do we do? In a nutshell, what we build is a hybrid storage system which comprises both solid state drives as well as low-cost SATA drives, or near-line SAS drives. So, low-cost hard disk drives combined with flash SSDs

Page 66

March 14, 2012

Industry Report

(solid state drive). And our data layout of our storage systems is very intelligent in saying all of our data resides on the hard disk drive portion of our system. We write all our data to the hard disk drive portion of our system. And what we also do is take the most frequently used or likely to be frequently used data and cache that in a solid state drive so all of our active or hot data sits in a cache which is made up of flash SSDs and 100% of the data lives on our low-cost drive tier. The real intelligence is that because we use flash inherently in our design, we can write extremely efficiently. We take all random I/O and write them as serial I/O to a disk tier. So even though we use low-cost hard disk drives, our write performance is dramatically faster than what traditionally disk drives have allowed you to do. So we write all our data to hard disk drives but our write performance is 10x of what you might expect from traditional hard disk drives because we have unique intellectual property in converting random I/O to sequential I/O. Thats how we get very fast write performance. Because we are able to cache the portion of our data into flash SSDs, we get very good read performance and so the overall system is good at both read and write performance. We also do some unique things. We compress data inline, something that no other storage system is able to do particularly for transactional data like databases and exchange. In our system all data is compressed and that gives us the ability to have more usable capacity for a given amount of raw capacity. And in fact that confers upon us both costeffective storage as well as high-performance storage. So those are some of the components that make us really effective when it comes to both performance and capacity. I also mentioned that in our system we are able to store primary data but whats unique about our system is that we can store hundreds of days of backups on the same system as well. And the simplest way for me to describe that is that we took all of the Data Domain approaches of deduplication and inline compression and instead of building those into a separate storage system, on the same system where your application data is being housed, we also store multiple days worth of backups that eliminate duplicate blocks and are compressed. There are customers of ours that have over a thousand snapshots. Over a thousand point in-time backups in the same system as the primary data itself. TE: The solution is pretty unique and I think its very compelling for end users. But do you see anyone close to your solution as far as competitors are concerned? Is there anyone working on it? Or do you think any of the established players could possibly work on this in the future? And when you compete, who do you see in the market? NS: In terms of who might be doing similar things, I would say there are lots of start-ups that are using flash. As you learned some of them are using flash very innovatively. But I think from a target market and from a focus of technology point of view, they are all focused on high-performance computing end of the market. Thats what they are going after. Whats unique about our approach is that we are targeting mainstream applications, mid-sized enterprises, mainstream applications where they are very conscious about storage economics. And thats really what we do effectively and I don't see many people doing what we do. In fact I don't know if there are companies targeting our customer base and using flash to address mainstream enterprise apps. So thats where I think we are somewhat unique what we are trying to do. Almost every one of the large companies can see the flash disruptions. Theyre all using flash in their systems. And while they can use flash to accelerate the portion of data that lives on flash theyre all taking some variation of the tiering approach and that gives you some performance benefit. Whats impossible for them to do is modify the disk layout of the storage system - so they end up adding flash on top of an existing disk layout. Theyre not modifying the way that the disks are being used and so theres only so far they can go in getting good performance efficiency or capacity efficiency out of their system and thats really what we are finding. We tend to be three to four times better on performance and at the same time were also delivering more usable capacity for a given amount of raw capacity. So we tend not to find anyone doing what were doing as well as were doing right now. TE: Who are you seeing when you go out and compete on a deal? NS: The most frequent competitor we see is Dell EqualLogic. They are 40% of the iSCSI market and its not at all unreasonable that we would see them. We see them in over 40-45% of our deals. Following Dell, it tends to be EMC - in particular, their VNX and VNXe product line is the one that we see most frequently. And then NetApp, with their FAS 2000 series product and even with their FAS3000 series product, is probably the third most frequently encountered competitor.

Page 67

March 14, 2012

Industry Report

TE: You mentioned that you're looking at the iSCSI storage market. Are there any specific segments or verticals where youve seen early traction? NS: There isnt a pattern in terms of vertical mix across our customers. We have over 250 customers now. And were adding customers at a rapid pace. But there isnt a vertical pattern. I would say what is common across our customer base is +80% of our deployments tend to be in server virtualization environments. Windows-centric applications tend to account for a majority of our deployments. These are often environments where our simplicity appeals to an IP generalist. These tend not to be large enterprise although we are starting to see more and more large enterprises. In particular, departmental-use cases are starting to come into play. But they tend to be more companies where really simplicity of deployment is a big factor. I mentioned what we do for performance capacity efficiency and what we do for data protection. The third aspect of our system that really appeals to our customers is that its dramatically simple. Dramatically simple to install and get going. So that tends to be a common pattern for our customers, less so a vertical mix. TE: And so can you share with us couple of key customer wins. And how you went about winning those? NS: Let me start with one thats pretty close to where we are. Its a customer called Plaxo. Plaxo was recently acquired by Comcast and they are an online address book hoster. They have a really large software development environment and they had a storage environment where one of their big challenges was build times for all of their software builds. They were facing two problems. First, the build times they were getting were fairly large and often it was causing loss of productivity for a lot of their software developers. More importantly, anytime they had a failure and they had a number of instances where they had to recover from backups, their developers would sort of be offline for hours at a time as they recovering data from backup tapes. That was the second big problem and even when they went to disk-based backup they still have to copy all the data back and that was a significant challenge for them. This happened to be a Dell deployment that they were contrasting with us against EqualLogic. The third issue that was also helpful was oftentimes they would have to create what I think of as copies of their development environment and make them available to various developers so the developers could work on their own sandbox. So all of these three were criteria were sort of what they were measuring their storage systems on. They have deployed us for some time now and theyve been a very happy customer. Their build times have gone up by a factor of two, if not three in many instances and have become faster. One of the best things that they found was we have a software function feature within our product line called zero copy cloning where you can use our instantaneous backups that I mentioned. Remember our backups take no time to create at all. But you can take any of these backup images and make them a writable backup image so when they had to create all these copies and provision those out to their developers, they could take any of these clones and say, Heres your version to work on. And it became a process of doing that in seconds. So that was a dramatic productivity enhancement for them. The way they deploy is they have an active-active replica with developers working on both systems. If they ever have any need to fail over they just move all of the developers on to one of our systems, and without any sort of downtime or interruption, they can continue to work. Thats been a customer where performance and our data protection solution mattered a lot. TE: And who are your key technology partners and how do you expect them to help you in your business? NS: The most significant partners for us now are VMware because we integrate deeply into the VMware environment. We work with all the VMware APIs. We have a virtual plug-in and so thats a very important partner for us. Another is Microsoft because 75% or so of our applications are deployed in Microsoft environments. We have deep integration with Microsoft backup APIs - things like Snapshot integration, Exchange, SQL and SharePoint. So thats another really important technology partner for us. Backup partners include CommVault, who has an API for managing snapshots. Then it becomes a whole host of other partners from a software application perspective. Then theres a bunch of components suppliers. SSD vendors like Intel that we have to work with that are essentially components in our storage system. TE: And when you look at the future, what is the vision for the business a few years from now? NS: When we look back as a company and as an executive team, we maniacally are focused on how we are driving growth and how we compare with some of the best storage companies that were started in the last decade. When we look at all the companies that really became successful over the last 10 years - companies like EqualLogic, companies like Data Domain. We sort of tracked their progress quarter by quarter. What was their revenue? What were their bookings? What was their customer acquisition pace? And one of the things we are proud of is in the first full year of operations,

Page 68

March 14, 2012

Industry Report

every quarter thus far, weve managed to stay ahead of the best of those storage startups. Weve grown at a pace faster than any of them grew in their first several quarters of being in operations. And so for us I think our vision is very straightforward. We see an opportunity to really penetrate a much more significant part of the market and grow rapidly so thats the first priority. Most of it is about scaling the business and driving growth. I think we have an opportunity to stay ahead of the curve that these companies established over the next two or three years and thats really all were focused on. TE: Can you give us a sense of your size at this time and your historical growth rate? NS: We were about 30 people as a company in November of 2010. We now have over 135 people. In the last 12 months weve more than quadrupled the size of the company. We have over 250 customers and we added about 60 customers last quarter. The pace of customer adoption is accelerating where we expect to add somewhere between 75 to a 100 customers this quarter. During this full year of operations we will have significantly gone ahead the growth rate of where EqualLogic was at in their first full year of operations. TE: Thats impressive. When you look out over the next couple of years, what do you see as your biggest challenge? NS: What Im very confident of is that weve now proven our value proposition to the customer. We know theres a really strong resonance with that. Our technology has been widely deployed and in a variety of environments to where we feel that the technology base is extremely solid and we are very confident of our roadmap. The biggest area of focus for us is really, How do we scale the business? We are growing in a place where we have quadrupled in size over the last 12 months and we think we have an opportunity to continue to grow at this pace over the next two to three years. So scaling is the only thing that I worry about. Its when you bring great talent on board, how do you make sure that you are not diluting the companys focus as you grow and how do you rapidly penetrate new markets. Weve just entered Europe; were not yet in Asia. And even within the US, our presence is there across the US but its a very thin presence so we are adding teams everywhere. Its really about scaling the organization as we scale the business and I would say that is the biggest priority for me. TE: Is there anything else you would like to share with our future interested clients? NS: The last comment I would make is that Ive been in the storage industry for a long time, nearly 15+ years now, and very rarely do you have a disruption that is so significant that it allows you to rewrite the rules of industry. So if you step back and you say for 50 years weve been living with sort of magnetic media i.e. tape for storing data. And then for 20 years, weve been living with electro-mechanical devices i.e. disk drives for storing data. And now we have a third type semiconductor-based media for storing data. I dont think its too much of an exaggeration to say that flash and SSDs require a rethink of how storage systems are deployed. And to me, I don't think Ive ever seen an opportunity where even multi-billion dollar product line and multi-billion dollar companies are vulnerable to a giant shift. And thats really what we saw as the opportunity when we started Nimble. And time will tell us if that proves right or wrong but Im certainly excited about what we can do here. TE: Thats great Suresh. I really appreciate the time youve taken to talk to us. I wish you all the best and look forward to tracking you in the future.

Page 69

March 14, 2012

Industry Report

Thomas Isakovich, CEO and Founder of Nimbus Data Systems Nimbus Data Systems, Inc. develops lP-based storage systems, software, and controllers. Its products include MySAN iSCSI Servers, iSCSI target software for Windows; Internet protocol Storage Controllers; and HALO Software, software for IP storage networking. Thomas Isakovich is CEO and founder of Nimbus Data Systems, Inc., where he oversees product strategy and development, sales, marketing, and manufacturing operations. Prior to Nimbus, Thomas was CEO and founder of TrueSAN Networks, where he raised over $34 million and led the development of storage virtualization and multi-vendor SAN management software. At Oracle spin-off Network Computer (now Liberate Technologies), Thomas oversaw product marketing for the companys line of thin-client workstations and server software. Here are excerpts of his interview with ThinkEquity LLC: Rajesh Ghai (ThinkEquity): Can you identify the key problem you're trying to solve in the data center? Thomas Isakovich (Nimbus Data Systems): The problem that we are solving with Nimbus is the I/O bottleneck and out of control power and data center costs in today's large scale enterprises and data center environments. TE: Why do you think existing solutions from the more established incumbents cannot address this problem? NDS: In short, the incumbents have designed their network storage products around disk technology. Whether it's the hardware or the software, they have a huge investment in conventional disk technology. Over the last 15 to 20 years there's been no other way except using disk technology. So they have invested huge R&D resources around disk, whether it's their file system and the way their file system technology behaves or, it's the hardware and the way the hardware operates, everything is engineered for disk. Flash technology is a complete 180 degrees away from disk. It's not limited by feed times and, it has completely different operating behavior than disk. So as you may expect, it requires completely new approaches when it comes to both the software and the hardware in order to enable the best possible use of flash technology. The incumbents simply are not focused on this. They see this as a sea change in the marketplace but, at the same time, they are presented with something of an innovator's dilemma where there are stuck in the same old way of doing things. So this creates an opportunity for a new company, like us, not encumbered by history of disk to take a clean slate approach and build a system from the ground up for flash, both software and hardware. TE: Great. And what are the key market trends that you are seeing in terms of what is driving demand for your solution? NDS: On one hand you have the constant desire for greater performance. Today's data centers must perform as fast as possible and must be as responsive as possible. New web infrastructure requirements are demanding more and more of a user experience and the response time of that user experience has become critical. So on one end, CPUs and networks and memory have scaled their performance dramatically but, disk has not. So that has created a huge I/O bottleneck with disk that solid state technology and flash technology, as we've developed, solves. At the same time, there is a struggle to reduce the cost of operating data centers. Companies do not have big budgets today to spend on power, cooling and rack space costs. Solid state and flash technology can those costs by up to 80% compared to conventional disk. Those financial savings are just as important as the performance gain. TE: Can you define the market you are targeting and what the size of that market could it potentially be, going forward? NDS: IDC estimates the performance enterprise storage market as about a $20 billion space. We're targeting that entire space. Conventionally, that space has been occupied by disk space storage arrays from the large server vendors and a handful of storage-specific providers. We are targeting that exact same space but, doing so with more modern hardware, i.e. flash and sold state technology and more modern software, i.e. the flash storage operating system that we have developed at Nimbus. TE: Ok, what are the key technology trends that you envision influencing demand within your target market then? NDS: With virtualization continues to expand in its adoption, which is driving the need for higher performance storage because the underpinning of virtualization is shared storage. The more you virtualize, the more burden you create on the shared storage infrastructure in the form of random I/O. Disk arrays are not well suited for random I/O. So as more organizations virtualize, there's a greater and greater demand for performance storage which favors a technology such as ours.

Page 70

March 14, 2012

Industry Report

While most of the virtualization done to date has been server-side virtualization, there is now also a significant movement underway to virtualize desktops. And similarly, desktops present another I/O demand in smaller organizations whether it's state and local governments or it's healthcare institutions and others that are adopting VDI or, virtual desktop infrastructure technology. That also is creating an I/O bottleneck for storage. At the same time, all other primary storage applications are a performance demand now that they've never seen before. Whether that's databases or e-discovery or, data warehousing or high performance computing applications or, scientific applications, everyone needs to go faster. And all those applications continue to grow and also fuel demand for archiving products. TE: Touching on that then, what specifically would you identify as the secret sauce for Nimbus? NDS: Nimbus is especially unique in the marketplace because it sits at the junction point of hardware and software for next generation storage. Storage systems are keyed by software content. Data protection software, data management software, data provisioning, all these functions are provided through software. Nimbus, over the past five and a half years has developed a comprehensive software operating system that we call HALO which provides a full suite of data management and data protection technology for flash based storage. The other vendors don't have that. They have disk-based storage tools. But not flash-based storage. The second thing is the hardware, of course. And Nimbus as well has developed its own hardware. We've developed our own flash modules that are extremely durable and very fast. And we've also developed the most scalable and versatile storage system on the market based on flash. We support protocols like Fibre Channel as well as 10 Gigabit Ethernet and InfiniBand. And we scale to hundreds of terabytes where would-be competitors are a mere fraction of that. So Nimbus is a unique player in that we have secret sauce in both our storage operating system and in the hardware of our products that we've merged together and provided as a complete solution out into the market. TE: When you're assessing the competitive landscape, which specific vendors do you feel are your nearest competitors and how do you anticipate the competition evolving over time? NDS: The competitors we principally run into are the large, incumbent storage vendors. You know companies like NetApp, on occasion, EMC, and then also due to the flurry of acquisitions, for example 3PAR by HP or, Compellent by Dell or, of XIV by IBM, those companies have also now become competitors of ours. But most typically it's NetApp and EMC. The reason Nimbus is successful goes back to the same argument that these companies design their products around disk. So they have nowhere near the performance characteristics of our products, they do not have the ease of use of our products, where we have designed our solution around flash and made it very simple to deploy. They do not have the same flash management technology that we do that ensures the durability of the system over an enterprise lifecycle. And finally, they don't have the same so called kind of rackanomics of our solution. We achieve far greater density with lower power consumption than our competitors. And we've done all this despite actually having a similar acquisition cost per terabyte. This is why Nimbus is really defying the conventional thinking. The perception in the marketplace is that flash, because it's a new technology, must cost 10 times the cost of disk. And that no one could make a cost-effective flash system. But Nimbus has done that. And we've been able to compete, toe to toe, with the incumbent disk vendors who are selling disks with an all-flash product at a comparable price per gig. And if you present an end user with that value proposition, that they could have all the speed and energy efficiency of flash and all the software features too, at a price point that's similar to what they're paying for old fashioned spinning disks, every end user goes and chooses flash. So we've been riding that success here over the last couple years and its particularly started to accelerate in the last nine months. TE: Taking that into account then, how do you sustain that advantage over the incumbents?

Page 71

March 14, 2012

Industry Report

NDS: Well, we need to continue to advance our hardware and our software technology. Hardware is very critical with flash. There's a general concept in the industry that off-the-shelf hardware should be good enough for any purpose, we don't believe that. We believe that a flash-based system, because it is so unique in its performance requirements, really does demand unique hardware. So Nimbus is continuing to invest in purpose-built hardware that maximizes the performance and the density possibilities of flash. At the same time, we need to continue to innovate on the software front. There's always more software you can do to provide greater data protection and data management and we're continuing down that path. And finally, we also need to provide even tighter integration into the applications and virtualization environments that are out there in use today, whether it's SQL server databases, Oracle databases or, VMware or, Citrix or, Microsoft Hyper-V virtual environments, the closer we can integrate at the application layer, the simpler we'll make management to the end user. Now the good news is we believe Nimbus has a good two and a half to three year head start on our competitors in flash hardware and software systems so we just need to keep our foot on the accelerator and we should be in great shape. TE: And then looking at your customer portfolio who, from a vertical or specific segment perspective, who is your target customer? NDS: Well this solution can be sold very broadly across many, many different vertical markets. It is usually the application that becomes the driver for this kind of a product. And with virtualization also now being broadly deployed, we can dovetail on that trend and sell into almost every vertical market. We have customers in healthcare, in education, in web infrastructure technology companies, financial services companies, energy companies, multimedia and Hollywood design studios and film companies. So it really runs the gamut. But where we have seen the most rapid uptake of our solutions is in the web infrastructure companies. Companies, for example, like eBay who is a significant customer of ours and has deployed hundreds of terabytes of our product, have done so because it enables them to accelerate how quickly their organization operates. And any performance advantage within the IT infrastructure, particularly for a web infrastructure company can directly translate into a better user experience. A better user experience translates into better eyeballs and more money for them. That's been the natural customer for us but it really is a broad solution. TE: Aside from e-Bay, any other key customer wins that you may want to cite? NDS: We have also significant wins in the U.S. government, that's another market that's pretty big for us. There are a variety of different DOD and DIA agencies that have deployed our technology for a variety of use cases. We also have achieved a number of other deployments. We'll be actually announcing a number of new customer wins here in early 2012. So I will hold off until we can officially those when they come out. TE: Can you provide a little further insight into your go to market model? NDS: Sure. We sell our products through channels as well as directly to strategic accounts. We believe the channel is the key asset for getting contact with the IT buyers. These are storage value-added resellers that have been selling storage products and actually have been eager to find the next thing. They've profited from prior start-up technology very nicely but then have also suffered when those start-up have been acquired and become part of the larger company then that's derailed the channel programs. They've been very eager to work with Nimbus and we've also been very, very eager to work with them and we provide a phenomenal program for the channel to profit from this tectonic change in data storage. TE: And then looking at demand in general, I know you touched on the general conditions surrounding that but how have you interpreted customer willingness to spend on a new solution such as yours versus an incumbent solution? NDS: I would say they are extraordinarily eager to consider, evaluate and, if the evaluation is successful, purchase Nimbus technology. They know that their incumbent supplier does not have a solution like this. The incumbent may be offering things that use a lot of same buzzwords to delay purchases and to maybe allow disks to survive another day in these performance-critical applications but, customers are have sharpened up and know that there's a limitation to what the incumbents can do and it's going to require a new company to deliver this kind of technology.

Page 72

March 14, 2012

Industry Report

Nimbus has established itself as the first mover in the space of having intelligent flash-based storage systems. We began shipping our products about two years ago. We have over 200 customer deployments. This is a production-hardened system that has stood the test of the toughest customers. And Nimbus, as well, is a profitable company. Companies can look at us and feel very financially secure in making a bid on this technology. Once we show them the price point and the capabilities of this technology, they are extremely eager. TE: You've touched on the current state of the industry and where the incumbents are versus your solution but what's your vision for the business say five years from now? NDS: Well I think within five years we can see all performance storage moving to flash. At a high level, the incumbents have been promoting a concept of tiering where end users will buy a little bit of solid state technology, a little bit of 15,000 RPM (revolutions per minute) technology, a little bit of 10,000 RPM disk technology and a little bit of SATA (serial advanced technology attachment) technology. Having all those tiers only makes sense if there's a huge price delta between those three tiers. If they all cost the same, everyone would buy the highest performing tier which, of course, is flash. We see over the next five years the cost of flash continuing to come down and effectively squeezing out 15K and 10K RPM disk from the equation. Those disk solutions will neither be faster enough nor cost-effective enough to really justify, we think within five years, a useful role in the enterprise. And when that occurs there will simply be two tiers in the data center, there'll be flash for everything that is performance-driven and, they'll be SATA for everything else. And, of course, you'll continue to have tape for archival and there may be some use cases of DRAM for that top 1% that's super, super performance where DRAM is at the very top of the pyramid. The market will simplify to flash and SATA and, Nimbus intends to be the number one player in that flash primary storage tier. TE: Could you give a sense of your size and historical growth rate? NDS: Well we've been growing very rapidly. This year we'll do approximately five times the revenue that we did in the prior year. So we are though still very much at the very, very beginning of the hockey stick. We are just scratching the surface of our potential and already growing very rapidly. I think though at the speed at which this industry is moving and Nimbus' competitive position in it, that we can definitely look toward reaching a revenue number that would make sense to consider becoming a public company somewhere in the 2013 timeframe. TE: Looking out over the next couple years, what do you see as the biggest challenge for the company aside, obviously, from the incumbents but in general for the company as well as dealing with the industry? NDS: Well I think our technology lead is sound. We started this company five and a half years ago, we have a significant software leadership position and hardware leadership position. I don't see new companies presenting a significant challenge to our technology position. Then the question becomes, how effectively can we get the message out to customers and, how broadly can we get it out so that they're aware of Nimbus? Those customers that we talk to about our technology are so impressed that they don't have to think very hard about the technology. If it does what we say it can do, it something that they're going to be buying and putting into their infrastructure. The number one challenge is simply scaling the sales and marketing force behind the company. I think if we do that and continue to push forward on technology, we should be in great shape. One thing I have noticed over the past year is the concern of the market overall around the durability of flash, that concern is going away. I remember a year and a half or, two years ago, when we embarked on this, some of the first question we received, was tell us about the durability concerns about flash because flash, unlike disk can wear out over time. How are you managing that and, so on. We still hear that concern. They want to understand how we solved that problem and that is a fantastic technology where we've solved that problem very completely. It's less of a concern amongst end users than it was a year ago, which helps. I think that signals a maturing market and customers are really embracing flash technology.

Page 73

March 14, 2012

Industry Report

TE: To wrap things up, I was wondering if there's anything else you'd like our institutional investors to know about Nimbus? NDS: Well, without being too dramatic we believe this shift to flash technology and flash succeeding disk as the replacement for primary storage truly does mean there's a $20 billion plus market that's developing here, going from zero just a couple of years ago. There is a phenomenal market opportunity here but, I think at the same time, it's critical that the institutional investors take a very close look at the underlying technology and the capability of the underlying product. Hearing the word flash is not enough. Just like hearing the word dot.com 10 years might bring you to the forefront, very quickly investors found out you know who the real leaders were and who were the wannabes from that group. I think the same thing, many companies are trying to profit on the notion of flash by mentioning the term but you really need to take a very careful look at what is the breadth of the software technology behind the solution because the software's key. All of the incumbents will say it's about software, software, software and they're right. You have to look very carefully at the software but, beyond that, you also need to look very carefully at the hardware. At the end of the day, 90% of the cost of goods that goes into this product is flash hardware. Whoever is figuring out the most efficient and highest performance system architecture for flash will win. And that combination is the key. I think if you do the homework in software and hardware to really get under the covers, you'll see that there are very, very few companies that have a cohesive story that will sell in the enterprise. Those are the companies to focus on.

Page 74

March 14, 2012

Industry Report

Scott Genereux, President and CEO, Nirvanix Scott Genereux is president and CEO at Nirvanix, a provider of enterprise-class cloud storage services. Prior to his CEO appointment in November 2010, Genereux was at QLogic where he served as senior vice president of worldwide sales and marketing. While at QLogic, Genereux expanded the company's presence in Fibre Channel and InfiniBand switching, forging successful design wins at tier-one OEMs such as Dell, EMC, HP, and IBM. Prior to QLogic, he was at DataDirect Networks (DDN), as senior vice president of worldwide sales, marketing and support. Subsequent to DDN, Genereux spent 18 years at Hitachi in strategic sales, marketing and support roles with increasing levels of responsibility. Here are excerpts of his interview with ThinkEquity LLC: Rajesh Ghai (TE): Can you explain to us the main problem that Nirvanix seeks to solve? Scott Genereux (SG): I think one of the biggest business issues that large enterprise customers are struggling with is how to manage and store data that is growing so fast and being generated at such a rapid rate that its becoming a nightmare for the IT department. Theyre constantly adding capacity to their installed storage systems and tape silos so they dont run out, migrating data every 3 years or so to move from one product generation to the next, or moving data to a different vendor platform altogether when products reach end of life. And for the type of data were talking aboutlarge unstructured content files--customers dont want to do data migrations every three years. Its costly. There are customers out there who are paying millions of dollars a year just to move data. When you factor in people and professional services its extremely, extremely expensive, and it doesnt add any value to the companys business at all. So how many times are you going to migrate from one tape silo to the next and how many times are you going to migrate from one storage vendor to the next? Some customers are finding that they can never catch updata is growing at the same time theyre migrating. They are doing it because the vendors are forcing them to do it. Theres companies out there today that constantly are EOLing their products, forcing customers to buy the next generation box they lashed together. NetApp now is EOLing their virtual tape product. EMCs Centera is headed for an EOL. What do customers do? How do they get off it? They are forced to move the data. You dont run into that problem with the cloud. Nirvanix reduces the cost and complexity of storage by eliminating the constant churn of data migrations. We automatically and seamlessly migrate data to new infrastructure on standard refresh cycles invisibly to the customer. The customer never notices it and is never involved. The key point here is that with Nirvanix, it will be the last data migration youll ever make, period. So I think thats a big piece of the problem we solve. On top of that with conventional storage acquisitions customers are paying for everything upfront regardless of whether they are actually using all of the capacity on day one and they always buy extra capacity so they dont run out. Of course, theres also monthly maintenance and support fees on top of that. With cloud storages pay by the drink model we are removing the heavy upfront costs for storage and eliminating CapEx. The cloud services pay by the drink model has become acceptable basically because you are paying for what you use, the ROI becomes extremely, extremely attractive to customers. Customers are no longer paying for RAID overhead which could be north of 40% in a single storage systemor metadata storage. Theres also the elimination of maintenance and tech refresh fees in addition to end of life concerns. As I said, Nirvanix takes care of infrastructure refreshes so theres no such thing as end of life in the cloud. And with this whole idea of being able to now obtain storage, availability and bandwidth resources on demand, theres no more waiting on purchase orders, vendor lead times, or dealing with procurement. I meet with a lot of companies and one of the things they constantly tell me is that the moment they need to buy storage in the conventional model, it could take three to four months to get through their procurement and purchase process entirely. With the cloud, that process goes away. So those are the main problem areas I would say we are addressing with our solutions. TE: Thats very interesting. I think you talked about a few market trends in your response. Can you also talk about what technology trends are making this possible today? SG: I think the obvious one is that business data continues to grow relentlessly. If you look at the current data growth forecasts, there will be nearly 8,000 exabytes of digital data created by 2015 and 90 percent of that is unstructured files, and thats what were focused on. Building complex tiered storage architectures is no longer practical. With the amount of data thats being created, you cannot continue to throw bodies at it. Youve got to do something differently. The massive data sizes, if you start thinking about it, with the onslaught of HD video and 3D video everywhere and photo libraries growing past the breaking point means companies have to look for new solutions to deal with new problems. The size of the data didnt exist three years ago, and I think thats the key. Companies want their IT resources available based on a

Page 75

March 14, 2012

Industry Report

model of consumption economics, and, you know, thats very important from an overall business management perspective. I think the need to reduce CapEx and shift to OpEx is a big trend across the board. Nearline storage is not cost effective for these burgeoning file archives, and tape storage cant satisfy compliance demands for rapid data access. If a company needs data quickly and it is on tape, it could take weeks or months to get it back. If its in the cloud, everything is tagged and indexed and you could access it at Internet search speeds and at a very compelling ROI. TE: And who is the target customer? Do you focus on any specific segments of verticals? SG: Our cloud storage services transcend vertical barriers. Every company in every vertical needs backup and archive. And of the new data being created by enterprise companies, roughly 60-70 percent is of the unstructured type. We have costumers in healthcare, financial services, media & entertainment, manufacturing and the information technology sector. Notable clients include Cisco and VMware for distributed content and collaboration. Data collaboration is a big use case for us, where companies can successfully integrate the cloud into their workflow. Other key customers include Cerner Healthcare, NBC Universal, Johnson & Johnson, Logitech, Comcast, Fox Sports and Warner Brothers. Over 1,200 companies worldwide are accessing our cloud storage network. A good example of cloud services transcending barriers is when we talk about rich media as a target market, a lot of times we think about the television and studios, the digital content creators, but in todays market, with the prevalence of iPads, iPhones, Droids and everything else, the fact is that every major corporation has rich media applications now. A recent example is a visit I made to one of our customers and one of the big discussions we got into was training videos. By the way, this company isnt a technology manufacturer or a media house, but they have 50 terabytes of video that theyd like to put out there so that people can download and access it around the world. This is going to be critical in every type of company. Across the board, we think theres opportunity in every vertical. TE: Thats great, and it seems like a very broad horizontal solution. Have you tried to estimate the size of the market that you could address? SG: Yes. Everybodys coming out with market data. We tend to look at the IDC numbers for the cloud storage space since theyre the most granular, and their latest estimates were for the public cloud storage market to be roughly $7 billion by 2014 and private cloud storage to be in excess of $11 billion by 2015. If you review their numbers, theres nearly about $20 billion in addressable market opportunity. Ive seen numbers higher than that by a factor of two though. What I tell people though is that even if you take the numbers and divide them in half, its still a gigantic opportunity in the cloud storage market no matter how you slice it. TE: Can you share with us some of your key customer wins since you joined the company? SG: Some of them weve obviously talked about. But Cerner Healthcare is a great example. Thats a multi-petabyte, private cloud. And thats an interesting one being a healthcare customer. Theyre going to use their cloud for internal purposes, but theyre also going to use their cloud to sell cloud storage to hospitals and doctors. And I think one of the things that cloud services is going to show the industry is that companies can shift from being purely consumers of IT to being providers of IT. So weve now enabled Cerner with their private cloud to go out and sell cloud storage services to all their customers and hospitals while EMC and HP and Oracle are in those same accounts trying to sell boxes. A year or two ago we would never have said that Cerner could be a storage company. Johnson & Johnson, Warner Brothers and Logitech are other new customer of ours. Weve had some phenomenal triple digit growth, and we expect it to continue in the coming quarters. TE: And can you talk about your competition? And then, more broadly, how do you see it evolving over the next few years? SG: It is necessary for those looking for cloud storage services to consider a vendor with flexible cloud storage deployment models. Some of our competitors only offer public cloud, others only offer a form of private cloud. We offer public, hybrid and privateall with usage-based pricing, all as a fully managed service. No one else does that. Not a reference architecture that might be supported and not a box with a cloud label on it, but a fully managed cloud storage service with 24x7 support and enterprise-level SLAs. Some of these public cloud-only vendors, such as Amazon S3, are focused on what I would classify as the developer space. They are having a very difficult time moving up into the enterprise space. So we see them as a competitor in the sense that while they offer a usage-based pricing model, their technology is more suited for the small and medium size business and the developer space. We dont see them, typically, in the Fortune 200, 300, offering the type of cloud capabilities that we do. EMC is another competitor, in its various forms, and I say that because personally I don't think EMC really has a cloud offering, even though if you look at their marketing it

Page 76

March 14, 2012

Industry Report

sounds like thats all they have. EMC goes in and pitches a barrage of unrelated products, from Isilon to Atmos to DataDomainthey pitch just about everything and anything and call it cloud. When I say EMC is a competitor, I really see us competing with EMC not necessarily because they have a cloud offering but because of their size and stature. EMC is a marketing machine and theyve done a really good job of painting the picture as the industry leader of what they call cloud. But, for them, cloud tends to be a piece of hardware or software and write me a check upfront Mr. Customer sales approach. And thats not what we call cloud, but, to be fair, I compete against that whole premise. Nirvanix is the most logical transition from the enterprise data center to the enterprise cloud. We focus on the companies that Amazon cant satisfy and that refuse to continue paying excessive upfront hardware fees to EMC and I think thats the key. So, my main competitor in the enterprise space is most likely whether or not they buy another box or shift to a fully managed enterprise cloud service. TE: You have talked about the hybrid cloud model. Can you talk about it and how that differentiates you from rest of the competition? I just want to understand how your service is unique. SG: We have our public cloud offering which enables you to put your data in one of our seven data centers around the world under the usage-based pricing model and leveraging our global namespace and billion plus sized object store. The other option we offer is what we call hybrid cloud, which has all the benefits of what you get in a public cloudusagebased pricing, you know, OpEx versus CapEx all the benefits that we already covered, but well put the cloud storage node in the customers environment, within their four walls. And its still connected to our public cloud, so they can access additional capacity, availability and bandwidth on demand. If they have any kind of security concerns, the first copy is in their location. Then the second or third or fourth copy can be anywhere in the world in our public cloud. And thats unique to us. Nobody else offers that today. Nobody else operates their own public cloud and federates that public cloud with privately deployed cloud storage nodes and manages the whole hybrid cloud federation as a service. This cloud federation gives you the best of both worlds. It gives you customer data on their floor and also data anywhere in the world, especially if you want to do data collaboration, disaster recovery, business continuity, and you have to get it offsite. And you got to get it offsite technically out of the region because weve seen regional disasters where power grids go down and having a data center down the street just doesnt work. So the hybrid cloud gives you the ability to do that. And again, with Nirvanix private cloud, its a private instantiation of a cloud storage node that is completely fenced off from our public cloud or any other cloud. Its your own private global namespace in your own environment. When our competitors say private cloud, once again theyre just selling hardware and some software. For hybrid and private, for the type of customers were talking about, they backup and archive a couple of hundred terabytes in a week or two. Were the only company out there that offers public, hybrid and private cloud storage services, with true usage-based pricing across the board. Nobody else does that, and that gives us a huge flexibility when talking to customers about building a cloud infrastructure that fits their business needs, not trying to force fit them into the only thing a vendor can do. TE: Sounds like a very compelling value proposition. Have you done any ROI analysis and can you share it if you have one? SG: Its interesting here in that customers want to pay by the drink for storage. They want data accessibility from anywhere around the world, not just a single data center. They want their data replicated across multiple geolocations, and they want to pick where it resides. So they want storage on demand, which is what we do, but they want their own hybrid clouds that facilitate access to a public cloud along with the right kind of security. Our customers find that we save them 70-80 percent over conventional storage system purchases. You have to keep in mind the maintenance, the tech refreshes, the data migrations that people are forced to undergo. Forrester recently did a TCO study and had similar findings. I was with a customer recently who told me that they spend more on EMCs remote replication software per year than they spend on EMC hardware. We dont do that. Thats part of our overall cost. So when you start thinking about how much does a customer pay for network costs and multiple copies, the actual capacity and the replication software and maintenance and the tech refreshes and the data migrations every three years, it becomes a pretty big savings. As we talked about, the primary competitive advantages for us include the ability to upload once, access anywhere with our global namespace. Were the only consistent storage on the Internet. Its self-healing, which I think is really important. For example a customer runs a job and the job fails. All of a sudden, a person has to get involved and orders a tape from Iron Mountain. It shows up hopefully two days later. If theyre lucky, the tape is readable. If its not readable, then he goes down this whole other path to recover the data if its at all possible. But that takes days. And, by the way, it could take weeks to get data back. With our cloud, he goes out there and accesses the data. Lets say theres two copies in our cloud and its in L.A. and New York. Lets say in our L.A. facility it cant read the data. The job doesnt fail. It picks it up in our New York facility where the secondary copy is, and the job runs and the customer has their data instantaneously. Its

Page 77

March 14, 2012

Industry Report

running, everythings great to the customer. The customer doesnt even know. The cool thing about our cloud is that the cloud goes back and says, why couldnt we read the L.A. facility? It determines whether or not there is an existing problem. And it basically takes the New York copy and fixes it for the L.A. copy. And so now were all synched up again. All that stuff is done without customer involvement. All of it is self-healing. All of it is done that way. This is what customers want for this type of data. When youre thinking that theres 8,000 exabytes of data thats being created, and its data that youre not going to access a lot, but when you do need it, you need it immediately. You dont want to throw a lot of people and time and money at it. You do for tier one data because the CEO or the CIO is calling, screaming at you. But, for this type of data, you want just to be as self-healing and requiring no intervention at all, and thats what weve created with our cloud. Thats what customers love, especially for tier three, tier four data. TE: Interesting. Given that cloud storage is a fairly new concept, I just wanted inquire into the level of interest as well as the customer willingness to spend compared to a year ago. How do you see that and how do you see customer interest evolving with regards to the solution? SG: Customers are aggressively shifting to the cloud. In fact if you look at our booking revenues, theyve doubled for the past two consecutive quarters. Whereas I think in 2009 and 2010 people were primarily doing evaluations while some early adopters were actually implementing, 2011 has seen a massive uptake in enterprise level cloud storage deployments. I think the uptake has been so significant that IBM has noticed this growth. IBM is now an OEM of our cloud storage technology and is integrating it with their cloud computing solution. Customers are shifting their buying intentions from the decade-old approach of conventional storage to enterprise cloud storage services. I probably spend 70 percent of my time talking to customers and prospects. I am always in front of the C-suite or VP of IT, and since I joined the company in the last year, 99 percent of the time every meeting I come out of, theres a follow-up meeting scheduled. There's a perfect storm starting. So I think thats a good indication that there's a lot of interest in what were doing, and I think the numbers and OEM partnerships speak for themselves also. TE: You spoke about IBM as a technology partner. Who else do you count as key technology partners and how do you expect them to help your business going forward? SG: IBM obviously as we talked about. We signed a five-year strategic OEM agreement with them so were pretty excited about that and the massive business opportunity that presents. Were excited about it because IBM is the number one IT services company in the world. They obviously have unparalleled customer relationships with the top companies around the globe. You know, were a service offering within their services company. It makes a lot of sense for what were doing. We also have a very strategic relationship with Symantec. NetBackup probably owns about 70 percent of the backup market for enterprise customers. Were the only cloud storage provider today integrated with NetBackup. We think it gives us a big advantage. Who knows what our relationship will be in the future but weve worked with them well for the last year. CommVault is a very strategic partner of ours and we are integrated with them. Theres also a company called Front Porch Digital. Theyre the Symantec of the rich media space. All three of these guys have written to our APIs so just like you have backup to tape, you have backup to the cloud. The same tools that they use to back up to any other platform are the same tools they use to back up to us. So its one click to the cloud. Cloud backends for business apps are the future and I really think youre going to start seeing that. We have reseller pacts with Dell and HP. This obviously allows access to their customer bases with our leading cloud storage technology. Those are some of our primary partnerships. TE: What is the vision for the business five years from now? SG: Our direction is to become an invisible part of our customers IT infrastructure and business processes through tight API integration and fully managed services; for cloud to be part of the core fabric of businesses around the world and enable them to thrive in their respective areas; for cloud storage to be as seamless as accessing everyday resources. You can expect cloud storage deployments in our customers at petabyte scale and exabyte scale. Nirvanixs experience with massive petabyte scale cloud storage implementation and enterprise-class production environments is unrivaled. Nirvanix manages real world enterprise customer environment with critical business data on a daily basis today. Nobody else has this experience in working with the Fortune 500 companies, placing petabytes of critical data into the cloud. Other cloud storage service providers dont know what they dont know because they havent done it yet, and I think thats the key. TE: I know that youre a private company and dont have to disclose your financials. However, can you give us an idea of your size and expectations of growth in the future?

Page 78

March 14, 2012

Industry Report

SG: Being a privately held company, what I can tell you is that Nirvanix has been posting triple digit consecutive quarterly bookings growth. We are doubling bookings every quarter in 2011. And we believe were taking material market share and posting strong, consistent growth. TE: Looking at the horizon over the next few years, what do you think are your biggest challenges? SG: I think we have to ensure that we stay more agile and flexible and continue to innovate as the large storage vendors begin to figure out their cloud storage strategies. I think its not advantageous for the current large storage vendors to offer cloud services and technology like what we have today. Its like going in your installed base and saying Im going to sell for 30-40 percent less revenue. And thats the problem they have. They are fearful of margin collapse should they shift from boxes to managed services. Over time Im sure theyre going to have to try to figure something out and its important for us just to make sure that we stay ahead of that and continue to innovate. And with the services that we are going to be rolling out in the next 6 to 12 months, I think is really going to show that. TE: Is there anything else that youd like our investor community to know about Nirvanix? SG: In closing, I think Nirvanix is positioned at the intersection of cloud innovation and market demand. We bring a very disruptive technology both from a financial model and a technology model to the market. Were at the forefront of what we think is a once in a generation paradigm shift. The leaders in the previous storage paradigm might not be the leaders in the new cloud storage paradigm. The landscape is changing and spending time defending a decades-old storage business model doesnt get vendors any closer to delivering real cloud business benefits to their clients. TE: Thanks for the time, Scott. We really appreciate your taking the time to talk to us.

Page 79

March 14, 2012

Industry Report

Dheeraj Pandey, Founder and CEO of Nutanix Nutanix, Inc. designs and develops converged compute and storage infrastructure for virtualized data centers. Its solutions include data center virtualization, disaster recovery, desktop virtualization, and testing and development. Prior to founding Nutanix, Dheeraj was VP of Engineering at Aster Data where he built the engineering team from ground up, overseeing the development of multiple releases of the product. Prior to Aster Data, Dheeraj was at Oracle where he managed development of the storage engine for Oracle Database and Oracle Exadata. Dheeraj holds a Masters in Computer Science from University of Texas at Austin, where he was a Graduate Fellow, and a Bachelor of Science in Computer Science from the Indian Institute of Technology (IIT). Here are excerpts from his interview with ThinkEquity LLC: Rajesh Ghai (ThinkEquity): Can you explain the key problem you are trying to solve in the data center? Dheeraj Pandey (Nutanix): The key problem that Nutanix is trying to solve is that of massive virtualization in existing enterprise data centers and virtualizations changing demands on infrastructure. We cannot retrofit an old architecture that was built in the 1990s physical environment on the growing demands of highly dynamic data center, which is what it is today with virtualization. The other key problem is that the network is the new bottleneck with the advent of solid state drives. On one hand you have server consolidation, which is causing storage access to be highly random and highly dense. And that is where we think we have a leg up on legacy network components such as Host Bus Adapters, switches and storage controllers built for spindles. We need a solution that will cut out the network from the middle. The solutions to the problem of big virtualization is no different than that of big data, which bring storage closer to compute and eliminate the network from the middle. TE: And why do you think existing solutions from the incumbent vendors cannot address this problem? NU: SANs (storage area networks) were designed in the mid-90s. Everything was done from the point of view of network storage, which was good enough for the workloads in the mid-90s and early 2000s, but the data center requirements have dramatically changed in the last five to 10 years. Existing storage vendors have zero to very minimal footprint in the server or hypervisor - -theyll just have to rewrite everything if they want to really have a footprint where storage and compute come together. Their solution assumes certain level of hardware and firmware assists which they cannot find in the server side commodity environment. TE: Whats the size of the addressable market you are trying to address through the solution? NU: Nutanix is not just yet another storage appliance. Its not a cheaper and faster NAS or SAN. We are bringing compute and storage together so we straddle both the server market as well as storage market. Together this total addressable market is in tens of billions. We believe that 90% of this will be virtualized by 2015, which means that this is a very large market. The other 10% that wont be virtualized, i.e., the vestiges of old physical infrastructure, is the new mainframe of this future decade. The real growth will be in virtualized data centers. TE: Great. What are the key technology trends that you are exploiting in order to build your solution? NU: One of the big trends that every enterprise small, medium or large -is looking at is the continued growth of private cloud. They are looking at how their data centers need to be more efficient, more automated, highly-elastic, and multitenant. The other one is consumerization of IT -tablets in the enterprise -and how that is causing virtual desktops to mushroom. For example, in a hospital where a doctor is carrying a tablet needs the virtual desktop to access all Windows legacy apps. Tablets, in conjunction with VDI (virtual desktop infrastructure), is a pretty big market opportunity. And the third one is enterprise flash. To shrink their storage footprint --data center real estate, power, cooling --customers are very seriously looking at this new silicon-based storage medium. That being said, until the economics of SSDs works out in $/GBs, they want the blended benefit of $/IOPS from SSDs and $/GBs from the old-style spindles. TE: Can you summarize your differentiation or your secret sauce? NU: The secret sauce is basically this Google file system like architecture that is hosted inside the hypervisor. Weve made the storage controller fully virtual, an arrangement in which storage servers sit next to a customers application

Page 80

March 14, 2012

Industry Report

servers. We have triggered this new form of vertical consolidation where application servers and storage servers can sit together in the same hardware. This is a very disruptive architecture. If you look at storage devices of the last five years, theyve increasingly been using x86 hardware, very much like the application servers of the last 10 years. There is a massive opportunity to share the motherboard, the fans, the power controllers, the chassis, the network, the management console, and the manpower to achieve integrated converged efficiencies between tiers of data center that were hitherto territorialized. The big challenge with converged infrastructure is scale-out: how do you scale to large numbers of machines without breaking the single-system-view goodness? You need a piece of technology that adheres to tenets such as no single point of failure, no single point of bottleneck, and scalable fault isolation --shared-nothing virtues that have made the big cloud companies scale. TE: Your solutions seem pretty unique and compelling. Who do you see as your nearest competitors and then, more broadly, do you see any of the companies that are not in the market right now developing a similar solution moving forward? NU: Convergence is a very attractive paradigm and pretty much any vendor who has a scale-out storage appliance will want to do this. If they have been using x86 hardware and if they know that their current platform can become a virtual controller inside of a hypervisor environment, theyd like to copy us. We have a couple of years of head start in understanding the complexities of a virtual platform, and how to make such an architecture reliable and highly available. Mere packaging is not going to help. There are some virtual storage appliances out there that can deliver branch office class SLAs in reliability and performance. Providing data center class SLAs is the hard part, and the biggest barrier to entry. Solutions that will be fast and reliable will win. Companies that survive on lipstick marketing will fall flat on their faces after the first wave of early adoption. This is a marathon that will be won by serious engineering teams that can put their arms around distributed systems, SSD reliability, and virtual platforms. TE: So this appears to be a pretty horizontal solution, I dont think it really lends itself to any specific segment or vertical, but can you talk about virtualized data centers or VDI where you are seeing initial traction? NU: VDI is a big one, as I mentioned, because of the recent requirements of tablets, universal access, compliance, and Windows refresh. But beyond that, almost all Microsoft workloads are virtualizable today: SharePoint, Exchange, SQL, and GP Dynamics. Microsoft has been preaching direct-attached storage and obviously VMware customers have been willing to virtualize anything and everything Microsoft. Its a great opportunity for us to come in and say look, Microsoft has been telling you to use direct-attached storage and we can do this in a very compact way in which multiple racks of spindles can be shrunk into a 2U form factor, with the same reliability and availability guarantees of a SAN. And that is powerful. TE: And can you share with us a couple of your key customer events? NU: Weve been seeing a lot of traction from federal, law firms, and hedge funds. There is a good reason for the traction in law firms and hedge funds. These are environments where the IT workforce is very small and they are constantly challenged on doing bigger and better things with technology because technology is a weapon its a competitive advantage they want to constantly lift the bar on. Weve started to see some early traction with customers who have big data (Hadoop) projects and have a VMware ELA. Licensing costs of VMware, thus, are not a problem, and virtualized Hadoop is ITs best friend. We are the only datacenter product in the market that is big-data-friendly compute and storage live in a single tier of machines in a massively scaled-out architecture. TE: Thats interesting. And what is your go to market model, are you going for the channel, are you going OEM route, or are you going direct? NU: We dont believe in OEMs. We believe that that model worked well in the 90s, but in todays world, a startup needs to tightly control its distribution and packaging strategy. You cant outsource packaging and distribution and then cross your fingers in a 3-yr wait-n-watch process hoping that someone elses BD and sales department will make you succeed because of their conviction in your product. Its a classic elephant-hunting sales model that fails more often than it succeeds.

Page 81

March 14, 2012

Industry Report

The channel --the regional VARs, the hungry tech-friendly mid-size VARs --is the only way for enterprise startups like ours to succeed. The solution that weve built is tailor-made for the mid-market enterprise. We have done 100% of our business through the channel up until now. This formula has always been known to work well --grow bottom-up with moderate sub100K deals for your meat-n-potato business. The key is to build a decent portfolio of early adopters and convert fencesitters with the help of a channel that leverages its trust with end customers. The laggards will then follow suit. TE: Great. And can you talk about your key technology and go to market partners and how exactly do you think they can help you moving forward? NU: VMware and Fusion-io are the key technology partners for obvious reasons. We leverage VMware as our platform, and we depend on a PCIe-based SSD for a stellar $/IOPS story. Fusion-io is the only vendor out there that has executed well in this space of PCIe-based flash memory, and has kept the economics friendly enough for the mid-market. TE: Great. If you look ahead, look at the future, what is the vision for the business in say five years from now? NU: Five years out, we envision enterprise datacenters to look not very different than what exist at Google, Facebook, and Amazon today --network storage will be scarce; compute and storage will run in the same set of machines as much as possible. This trend is already pretty evident in the big data world. And that is how Pervasive Virtualization will look like in the coming years. TE: When you look over the next couple of years, what do you see as the biggest potential challenge for your company? NU: I think the biggest potential challenge is the macro environment. At some level, this is a complete unknown. If you look at whats happening in Europe and how its affecting 2012 and beyond that, this looks like a revisit to 2008 and 2009 already. That being said, I think for a startup like ours flush with money, we can use it to our advantage. Its instructive to note the transformation of recession-friendly products such as VMware between 2008 and 2011. The product (pre-2008) was built and marketed for test and development environments. Then as the recession grew deep and wide, VMware began to aggressively penetrate data centers and by 2011, people are saying sky is the limit for virtualization. Something similar happened with Microsoft when the Web 1.0 bubble burst and the economy went into recession. By 2004, Windows dominated 40% of the data center, and SQL Server had become pervasive in the midmarket. Nutanix is perfectly poised to convert this sluggish environment to its advantage with its cash flow, technology, and sales/marketing hygiene (who to sell to). TE: So to wrap up is there anything else that you would like our institutional and investor clients to know about Nutanix? NU: We have encircled ourselves with advisors, board members, and employees who have seen -and were a critical part of --the formative years of channel-friendly businesses such as Data Domain, VMware, Riverbed, Palo Alto Networks, and Aruba Networks. Look out for Nutanix in the coming 24 months. TE: Thats an excellent overview of Nutanix for our clients, Dheeraj. I appreciate your time and good luck and I look forward to tracking you in the future.

Page 82

March 14, 2012

Industry Report

Scott Dietzen, CEO of Pure Storage Pure Storage provides enterprise storage solutions. Its products enhance random I/O-intensive applications, such as server virtualization, desktop virtualization (VDI), database (OLTP, rich analytics/OLAP, SQL, NoSQL), and cloud computing. Scott Dietzen was most recently president and CTO of Zimbra, a vendor of open source messaging and collaboration software. Zimbra (now part of VMware) was a pioneer of the Web 2.0 and Open Core paradigms. Zimbra was originally acquired by Yahoo!, where Dietzen served as interim SVP of Communications and Communities. Prior to Zimbra, Dietzen was CTO of BEA Systems, where he helped craft the technology and business strategy for WebLogic that drove BEA from $61m in annual revenues prior to the WebLogic acquisition to over $1B. Dietzen came to BEA in 1998 via the acquisition of WebLogic, an innovator in Java and web application server technology. He also currently serves on the board of big-data trailblazer Cloudera. Here are excerpts from his interview with ThinkEquity LLC: Rajesh Ghai (ThinkEquity): Can you explain the key problem you are trying to solve in the data center? Scott Dietzen (Pure Storage): Mechanical, disk-based storage is slow, power- and space-hungry, and expensive. From the perspective of a CPU doing the random I/O required by virtualization and database workloads, todays hard drives appear slower than tape did 15 years ago. This is because CPUs get faster, while random I/O disk only gets slower as we pack exponentially more data onto a device whose mechanical disk arm performance is flat. It is as if we made a bus 1000x longer, but didnt add any new doors. Mechanical storage also typically consumes more than 40% of a data centers power budget. And its still expensive relative to the value delivered, with shared disk storage generally costing the enterprise more than $5/GB, typically with much of that capacity unutilized as end users short stroke their disks in order to hit performance targets. We are convinced that the continuing evolution in data center servers and networks has left mechanical storage behind, and that end users can now get far more for their money with solid-state storage. TE: Why cant existing solutions from the larger more established vendors address this problem? PS: By purpose-building a storage solution for Flash memory, we are delivering a product that is 10x faster (both in terms of latency and throughput), 10x more space and power efficient, and at a price point that is competitive with that of performance disk. Retrofitting Flash into a traditional disk array costs well more and delivers well less than the Pure Storage approach. The challenge is that the controller software of the disk vendors has been customized over many years to optimize to the idiosyncrasies of mechanical disk, but solid-state Flash is a profoundly different media, and requires dramatically different optimizations. When Data Domain got started (building a backup solution for disk rather than tape), few argued that they should start with the software from a physical tape library, and yet thats precisely what the incumbent mechanical storage vendors are doing. TE: What are some other key market trends that you see influencing demand in your target market? PS: In our view, mechanical storage will be superseded by solid-state for most performance intensive workloads over the next decade, while capacity workloads like backup and archiving will clearly remain the domain of mechanical storage for many years to come. Cloud (a.k.a. elastic computing) is also going to drive the storage market going forward. The interesting dynamic with the public cloud is that there has been a tendency to use server-local disk because it is more cost effective---Google, Facebook, Amazon have all pioneered this practice. However, for Flash it turns out that server-local deployment is substantially more expensive than deployment within shared storage appliances or arrays. That is because in a shared appliance you can more carefully manage wear leveling and the periodic refresh of aging data, allowing the use of most cost-effective MLC Flash. More importantly, a shared appliance accommodates far more effective data reduction (dedupe & compression), allowing the end-user to store 5-10x more data on the same capacity of Flash. So perhaps surprisingly, the economics for Flash in the cloud will actually run in reverse of those for hard drives. This is a great trend for proposebuilt Flash arrays like Pures.

Page 83

March 14, 2012

Industry Report

TE: Can you give some sense about the potential size of your addressable market? PS: The performance disk market is approximately $15B and 10 exabytes of capacity a year. Over time, we expect much of this market to move to solid-state Flash. There are other interesting solid-state technologies such as Phase Change Memory (PCM), but Flash has the advantage of a 20 exabytes a year consumer market in smartphones, tablets, laptops, cameras, etc. that has driven down costs for the data center. TE: What are the key technology trends that you see influencing demand in your target market? PS: Just listing them out. The transition from mechanical to solid-state; the emergence of the public the private cloud; we are seeing data center capacity expansion gated by power along with virtualization/consolidation driving up random I/O. Additionally, there is a lack of OLTP (online transaction processing) latency tolerance driving faster storage and user demand for more instantaneous response time---think Google instant search. Finally, more complex database analytics result in users demanding faster storage and a resulting blending of OLTP and analytic database workloads demanding faster storage. TE: Can you provide some more detail as to how your solution addresses the problem? PS: To make Flash effective in the data center you have to manage its idiosyncrasies. Unlike disk, Flash is random access---more like memory than tape---that necessarily entails a different data layout. Unlike disk, Flash is asymmetric in that it is more expensive to write than to read, which means you want to write the Flash very carefully. Unlike disk, signal quality on Flash degrades with use over time and should be automatically refreshed to maximize longevity at the cheapest price point. And unlike disk, Flash is more expensive than CPU, so it makes sense to spend some CPU in order to deliver high-performance inline data reduction (dedupe & compression) to get the cost of Flash down to where it is competitive with mechanical disk. The same data reduction techniques simple do not make sense on hard drives. Consider that deduplication entails replacing a contiguous block of data with pointers to repeat segments---that is, replacing contiguous I/Os with random I/Os for which disk is >95% inefficient which results in lots of seeking and rotating, for minimal transfer times. TE: What specifically is your secret sauce? PS: Traditional storage solutions have been specialized over years to take best advantage of the underlying mechanical storage media (a.k.a. hard drives). We are doing the same thing for Flash memory. Flash is deserving of the same customized software stack to maximize its performance and reliability that disk has enjoyed for decades. Pure Storage so carefully manages the Flash in our array that we can take Flash that has been written well past its warranty and whose performance had degraded commensurately, and yet when we insert it into our product, it behaves like a pristine drive. We of course do not allow this to happen in our customer configurations, but it is testimony to what careful Flash management can achieve. However, in other respects we are more like a traditional enterprise array. We have a multi-controller architecture for scalability as opposed to an individual appliance wherein each controller must be managed separately. We provide true high availability in that our hardware is redundant, and our software is designed to automatically heal around any failures that crop up. Traditional array feature sets are going to be essential to long-term success in solid state storage---end users are rightfully unwilling to give up snapshots, replication, and so on. So we are convinced the winning solid state storage vendors will primarily be software companies, as our disk brethren were before us. Most of all, Pure Storages secret sauce is in delivering very high performance data reduction algorithms relative to deduplication and compression. Were averaging >5x for database and >10x for virtualization workloads, and thats not counting savings from thin provisioning. Prior to Pure Storage, there simply were not similarly effective data deduplication techniques that could operate inline and with sub millisecond latency. TE: Who do you see as your nearest competitors and then more broadly how do you see competition evolving? Do you typically replace a solution from one of your competitors? How intense is the competition right now? PS: Storage is a perpetual green field market in that organizations data footprints are typically growing at double digit rates, and their existing storage arrays have a 3-5 year lifecycle. So if a customers footprint is growing 20% a year, and they are aging out 20% of their current storage annually, thats 40% of their current aggregate footprint they need to buy each year.

Page 84

March 14, 2012

Industry Report

Our primary competition today is coming from the incumbent mechanical storage vendors. That will continue, although we also expect to see traction for other solid-state storage companies as well. We definitely expect the competition to be ferocious and the market to consolidate. At present, there is activity in three different categories. 1) There are Flash storage appliance vendors that are focused on establishing a new Tier 0 of data center storage. These solutions are like race cars in that they are very fast, but also tend to be expensive and lack the features necessary for routine use. 2) There are vendors focused on aggregating Flash and disk within single appliances for primary and secondary/backup storage. These products typically offer relatively lower price points, and tend to appeal to the small business and small enterprise markets. Larger data centers tend to want to avoid putting backups on the same storage platform as primary storage to reduced risk and increased specialization. And finally, there are vendors in the middle like Pure Storage that are focused on mainstream Tier 1 performance storage. The primary competition in this category will be between the existing mechanical disk vendors with retrofitted Flash and the new solid-state vendors offering data reduction as a means to bring Flash in line with hard drive cost. TE: How do you sustain your competitive advantage? PS: We expect the hardware for solid-state storage to commoditize, much as the hardware for mechanical storage did. That means the differentiation will primarily be in software going forward. The winning solution will be the one that continues to deliver a faster, simpler, more reliable, and more cost effective platform than our competitors. Pure Storage was a first mover in both marrying all-Flash design to a traditional enterprise array feature set as well as in high-performance data reduction. We will certainly do our best to maintain the leadership position we have carved out with our innovations thus far. TE: Who is your target customer any specific segments or verticals? PS: Enterprise data centers and public cloud providers. Currently our customers include virtually every vertical, including financial services, telecom providers, logistics, energy, pharmaceuticals and agribusiness, legal, government, and so on. TE: Please share with us a couple of your key customer wins and how those were won. PS: eMeter, Fenwick & West, and MatterSight are all examples of companies that have elected to go public with their use of Pure Storage. In each case, it was our combination of performance, power & space efficiency, and economics that won the customer over. Data reduction is core to our economic value proposition. We estimate that about 5:1 dedupe and compression make Pure Storage break even with performance disk on cost and, of course, 10x better on performance and power. Each of these customers exceeded that 5:1 target for their workloads, making our value proposition a relative no brainer from their perspective. TE: What is your go-to-market model? PS: We are primarily going to market through VAR channel partners, although we are working with our VARs to touch every one of our early customers directly to ensure their success and so that we can learn as much as possible about how our product performs in their environments. TE: Who are your key technology and go-to-market partners and how exactly do you expect them to help you in your business? PS: Samsung and STEC are both critical hardware technology suppliers to Pure Storage. We also rely upon and contribute back to a variety of open source software infrastructure such as Linux. Those are the key upstream providers. Downstream the focus is on VAR channel partners that are helping our end users replace mechanical storage with solidstate, and doing so within their existing budgets. TE: At this point of time, how are demand trends and customer willingness to spend on a new solution such as yours? PS: We have far more interest than we can accommodate right now, and have had the challenge of trying to pick among candidate end users in terms of finding the most interesting workloads and highest quality feedback. No doubt it is still very early stage for us, but we literally see nothing constraining broader deployment of Pure Storages products other than our ability to continue to harden our technology and grow the business organically as quickly as possible.

Page 85

March 14, 2012

Industry Report

TE: What is your vision for the business, say five years from now? PS: We aspire to be the largest vendor of shared, all solid-state storage. TE: From a product roadmap perspective, what new product offerings or features and functionality that youre working on what investors should be aware of? PS: As a private company, we are not publishing our product roadmap. TE: Can you give us a sense of your size and historical growth rate? What are your revenue projections for the future? PS: As a private company, we are not publishing our financials. TE: When you look out over the next couple years, what do you see as the biggest potential challenge for your company? PS: This will be a ferociously competitive market. End users will need to do their due diligence, but they are also in a position to get much more for their money if they shop wisely. TE: Scott, thank you for your time and good luck.

Page 86

March 14, 2012

Industry Report

Dave Wright, CEO and Founder of SolidFire SolidFire offers an integrated hardware/software storage solution that runs on physical servers to deliver a highly scalable block storage. SolidFire also focuses on solid-state drives, which provide performance advantages, and low latency response times across highly varied workloads. Dave Wright helped start GameSpy Industries, a leader in online videogame media, technology, and software. While at GameSpy, Dave led the team that created a back-end infrastructure powering thousands of games and millions of gamers. GameSpy merged with IGN Entertainment in 2004 to create one of the largest Internet gaming & entertainment media companies. Dave served as Chief Architect for IGN and lead technology integration with FIM / MySpace after IGN was acquired by NewsCorp in 2005. In 2007 Dave founded Jungle Disk, a pioneer and early leader in cloud-based storage and backup solutions for consumers and businesses. Jungle Disk was acquired by cloud provider Rackspace in 2008 and Dave worked closely with the Rackspace Cloud division to build a cloud platform supporting tens of thousands of customers. He left Rackspace to start SolidFire in December 2009. Here are excerpts from his interview with ThinkEquity LLC: Rajesh Ghai (ThinkEquity): Can you explain the key problem you try to solve in the datacenter at SolidFire? Dave Wright (SolidFire): SolidFire is focused on solving the challenge of primary storage in large-scale cloud computing datacenters - essentially in large, cloud datacenters, that is datacenters where you've got thousands to tens of thousands of virtual machines in a cloud-type environment. We are solely focused on the primary storage challenge associated with cloud computing and particularly the challenges related to performance, efficiency and management. TE: And why do you think existing solutions from larger, more established vendors are unable to address this problem? SF: I think the simplest explanation is that those solutions were designed for different environments and different problem sets. If you look at the enterprise storage technologies that are available for primary storage from the major vendors today, they are really focused on the kind of enterprise storage challenges that were predominant for the past five or 10 years - running small numbers of high performance applications on SAN-type (storage area network) storage and with a kind of high-touch, high-level of administration with dedicated storage administrators. In a cloud environment, we are moving into an environment where you now, rather than just having a few dozen applications on your SAN, have thousands or tens of thousands of applications that need primary network storage and the capability to scale that environment, to manage that environment and to automate operations within that environment. The requirements around that are completely different than what they were in a kind of traditional enterprise storage setting and while they've tried to develop and market their products into this segment, into this opportunity; at the end of the day, those products just were not designed for the challenges that large-scale cloud computing poses. TE: You touched a little bit on this but can you summarize what the key market trends that are influencing demand for your kind of a solution? SF: Really there are two large trends that are underway right now that are driving the need for this technology. The first is if you look at the kind of service provider and hosting space - the transformation that is going on in that space from a model where there used to be service providers that had dedicated or managed hosting environments and there were service providers that had low-end, shared hosting environments into a converged, cloud-hosted model that uses virtualization to allow dense hosting of lots of tenants, lots of applications in a very dense storage and compute footprint in these large, public, hosted clouds. Places like Amazon, Rack Space and SoftLayer that are building these large infrastructures of service-type cloud environments. That is really driving the need for new types of technologies to run these large, multi-tenant public clouds. The second challenge is if you look at the enterprise and particularly the high end of the enterprise, the Fortune 500 that own and operate their own datacenters. They are striving to make their datacenters more efficient, more reliable, and more cloudlike and really become more like a service provider to their internal organizations and they are adopting a lot of the same strategies, practices and technologies that the service providers are to virtualize, consolidate and manage their growing computing footprints. Those two trends together lead to a world going forward where large-scale infrastructure type computing in cloud environments is really how computing is managed for both large, public companies as well as smaller and medium sized

Page 87

March 14, 2012

Industry Report

businesses that are hosting their environments in public clouds. Really the entire trend of where enterprise IT is going at the small and large scale feeds into this. TE: You said that you are targeting the primary storage market and you talked about the public cloud market, you talked about the enterprise private cloud. Can you say that this is a very broad solution that kind of addresses the primary storage market and if you can give an indication of its size, that would be helpful. SF: I think that it is something that is primarily focused on large scale cloud environments which, right now, you have got a lot of deployment of those in service providers. And you have the start of deployment of large scale clouds and enterprises as well. But, I think, going forward, what you will find is that the percentage of applications and the percentage of computing that takes place in large scale clouds versus smaller clouds or smaller, one-off datacenters or one-off server rooms. The ratio is going to shift so that more and more of the computing and more and more of the storage exists in larger cloud environments either in an enterprise cloud or in a public cloud. I don't have specific numbers within the market right now to share on the size of the kind of public cloud and large private cloud market but it is definitely a several billion dollar market today and I think it is going to grow to include the majority of the enterprise storage market over time. TE: And what are the key technologies that you are leveraging to build a solution? SF: There are a couple of key technologies in what we're building that are unique to our solution but also allow us to serve some of these challenges for this kind of large cloud market. First, is the use of solid state storage and flash memory technology. Our system is all solid state. We do not use any spinning discs in it whatsoever and that allows us to drive significant improvements in performance, density, power utilization over the disk-based storage systems. Moreover, in combination with some of our unique IP around performance management, it allows us to give guaranteed performance and guaranteed quality of service to thousands of applications at the same time. Some other key technology drivers that are behind our solution are things like the ubiquity of 10 gigabyte Ethernet. Our system is a clustered storage solution that connects both with itself and with the compute nodes on the network over 10 gigabyte Ethernet using iSCSI (small computer system interface). It also leverages the industry standard of Intel-based form factors for the actual server chassis so no proprietary hardware thats part of the solution. And really the last big trend that is starting to emerge that we have really leveraged in our architecture is data efficiency and data reduction technologies, the ability to use compression and deduplication and thin provisioning but do them in such a way that it is designed into the architecture from the beginning and can be done essentially in real time without performance impact. That is really driving down the cost of the solution as well as increasing the density, increasing the power efficiency. TE: And what specifically would you define as your secret sauce and how are you leveraging all these technologies? SF: There are a couple of key things that we do very well. One is this capability to build a scalable clustered primary storage system, the ability for an enterprise or source provider to start with a certain footprint and add both more capacity and more performance in a linear fashion to this cluster up to a very, very large size, enabling the system to be very scalable as these cloud environments continue to grow over time. The second big kind of piece of our technology in IP is the data reduction technologies that we talked about, the ability to run those in real time without performance impact, increase the effective capacity of the system and make it cost competitive even on a dollar per gigabyte basis with diskbased system while having significant benefits in terms of the cost structure operationally. And the last piece and this is probably the piece we are best known for is this capability we call performance virtualization. And it is basically the idea that you can separate performance from capacity in the storage system, have a separate pool of capacity and a separate pool of performance. And for every single volume or application that is provisioned storage for in the system, you can dial in exactly how much capacity and exactly how much performance it needs and then you can turn those dials over time to give more or less performance, more or less capacity as needed but do so in a way that you can guarantee that performance of that application irrespective of what else is going on in the system and that you can adjust that without having to physically move or tier data to different levels of storage, different types of disks or different storage environments. It is all managed internally within our system.

Page 88

March 14, 2012

Industry Report

TE: That sounds very compelling and unique. If you look at your radar right now, do you see anyone kind of building something similar or trying to build something similar in the future? SF: We definitely see bits and pieces of our strategy in other solutions so you are starting to see other all solid state solutions out there from vendors like Avere, Violin Memory, Texas Memory Systems and Nimbus. So you're starting to see other systems that are made up of all solid-state storage but generally, those systems do not have the scalability or management components of what we do and they may or may not have the data reduction features. In some cases they do and some cases they don't. But in general, really what they're missing is the scale aspect and really the focus on largescale, multi-tenant environments that we have. We really just do not see that in any of the emerging flash vendors. Now, of course, among the larger established players, you know, there is certainly a lot of large-scale, primary storage systems, clustered storage systems that are available from the likes of EMC and 3Par and the like. But again, by in large, they were designed for a different environment with different needs and not for a large-scale, multi-tenant cloud. They dont effectively leverage Flash technology. They don't effectively leverage data reduction technology and so they end up being very expensive, very complex, very difficult to manage and just not at all suitable for the large-scale cloud. And while it would be nice to think that they are going to innovate in those areas, they have huge investments in these platforms and we really think that to really address these challenges, you need to re-think the architecture of the storage system. TE: Can you share with us a couple of key customer wins and how you went about winning those? SF: We haven't discussed publicly any of our customers. We are in a kind of a beta phase of the product right now. Where we have the system deployed is at several service providers around the U.S. We are going to be adding more service providers to that pool in the first quarter of next year. And we haven't announced any of those yet. But we've got a pretty extensive number of service providers that we are working with either in early assessment phase or just prior to that phase. And the desire that we are seeing from these guys is really all around that performance, efficiency and management story where they just are not getting that from existing storage solutions. TE: And how do you plan to go to market for this opportunity? SF: We are initially focused on large-scale public clouds and we are going direct to those service providers, to the guys that are powering those large clouds. TE: OK. And who are your key technology or go to market partners at this point of time? SF: We've got a couple of different technology partners that we have not yet announced. We've got some relationships that we have announced with Arista and Xsigo but we're still probably in the early phases. A lot of what we are doing is integration with not just other kind of network and storage hardware solutions but also integration with cloud and cloud management stacks and so we are doing work with OpenStack. We're doing work with VMware and other systems in order to really integrate our storage at a deep level with those environments to make it very seamless to provision and manage storage with SolidFire. TE: When you look ahead, what is your vision for the business five years from now? SF: Really what we see is that the fundamental technology we built has a pretty good road map ahead of it in terms of being able to expand to manage more and more parts of the primary storage spectrum. Now there is certainly a need for other types of storage including secondary storage. But that is really not our focus. Our goal over the next five years is to make our technology ubiquitous in public and private clouds that are of certain scale. We are not interested in the SMB and mid-sized market. We're really looking at large enterprise and public clouds. But we really want to make our technology ubiquitous in that market over time and try to be the best possible technology for primary storage in the cloud. TE: What do you see as the biggest challenges for a company like yours at this point of time? SF: I think the challenges are the same for any company that is delivering their first product; we've got to get the product to market at the right time, at the right price and with the reliability and functionality the customers expect. I think we have done a good job on a number of those items. There are a number of them that we're still working on in order to bring the product to market. But the demand is definitely there; the interest is extremely strong and we just need to deliver the product, make the customers ecstatic about it and keep going from there. TE: And to wrap up, is there anything else you would like our institutional investor clients to know?

Page 89

March 14, 2012

Industry Report

SF: I think from our perspective SolidFire is riding a couple of different waves that are about to converge and amplify against each other and certainly one of those is solid state and the solid state storage wave. The second is cloud computing and the move from siloed computing into cloud computing which is overtaking every area of business right now. I think this another huge trend that we're leveraging. For us, what we are building here is not just a better way to do storage that's kind of deeper, better and faster. The cloud is the evolution of the way that IT infrastructure will be managed. We really think that SolidFire and the architecture we have developed is the evolution of primary storage and how it will be managed, really a generation or two ahead of what you see from enterprise storage vendors in the market today. TE: Thanks for your time and good luck.

Page 90

March 14, 2012

Industry Report

Ursheet Parikh, CEO and Founder of StorSimple StorSimple, Inc. provides cloud integrated enterprise storage that combine the function of primary, backup, archival and disaster recovery into a single appliance that is seamlessly integrated into cloud storage services such as Amazon S3 and Windows Azure. The solution is the only public cloud integrated enterprise storage that is certified by Microsoft and VMware. Prior to StorSimple, Mr. Parikh was the Director of product management for Ciscos WAN optimization and application acceleration business unit, building the Wide Area Application Services (WAAS) product from concept to over $250 million per year in revenues. Mr. Parikh came into Cisco through the acquisition of Andiamo Systems, where he held a product management role. Prior to that, he held leadership positions at Kuokoa Networks, a file system virtualization company and at Jalva Media Inc, a digital asset management software and outsourcing company. Mr. Parikh started his career in enterprise IT at Deloitte Consulting where he worked as analyst, technical architect and project leader for a range of data center, networking and applications IT infrastructure projects across multiple industries. Here are excerpts of his interview with ThinkEquity LLC: Rajesh Ghai (ThinkEquity): As we kick off 2012, I wanted to ask you, what was the most seminal thing that you saw in 2011? Ursheet Parikh (StorSimple): I think in 2010 we had the introduction of the iPad and that has fundamentally changed client computing and is driving so much more innovation on the desktop and the mobile device management side. I think something similar happened in 2011. I dont think that has been appreciated enough by most enterprises. And thats really iCloud. What Apple did with iCloud is that they basically made the concept of the cloud very accessible. They moved the center of gravity away from PCs, which were the resident for peoples information and moved all of that data into the cloud. Whats amazing about it is the agility with which they delivered the service. And that is where we think a lot of the future of enterprise IT is headed. They have delivered this service of incredible scale at a really low cost and have really built it out as a meta service because the core infrastructure that underlies iCloud really comes from the Amazon storage service and the Azure storage service. You end up having this really interesting point where tens of terabytes of capacity have to be added at incredibly low costs and build these services really fast so that they can focus on the core competency around user experience. They just couldnt find any of the traditional vendors up to the task and had to rely on the public cloud services, and in some cases, a competitor like Amazon to make it happen. TE: If you look at StorSimple, can you tell us what the key problem youre trying to solve in the data center is? SS: The crux of what StorSimple ends up delivering is best described as iCloud for the enterprise. If you see how enterprise data centers are set up, computers are highly optimized. In the past decade, with what VMware has done on the compute side, your compute utilization has gone from 10% to close to 70, 80% on the servers. And the footprint, as a result, has also shrunk significantly. You now have a problem where, when it comes to the enterprise data center, all the different storage products run primary data such as disk-based backup products, tape infrastructure archive, extra copies for disaster recovery. What makes up this full range of storage systems really now accounts for almost two-thirds of the footprint in most enterprise data centers. What StorSimple really does is actually collapses these different functions of different storage devices into a single appliance that is seamlessly integrated into the cloud. As a result of which it actually ends up bringing the economics of cloud to virtualized enterprise data centers and allows most of these enterprises to tap into billions of dollars that are invested in data centers by the likes of Amazon, Microsoft, AT&T, Rackspace, and other big cloud computing providers. TE: OK, thats interesting. If you look at the more established storage vendors out there, why do you think theyre not looking at the solution or why dont the existing solutions not address this problem? SS: If you look at StorSimples customers right now, all of our customers who have purchased us have either entirely replaced for a specific application or a department, or have augmented for a specific application or a department in Fortune 1000 accounts existing storage solutions from the likes of the major vendors that you can name, NetApp, EMC, IBM, HP, Hitachi, Dell. Fundamentally, what is happening is that when they are making a decision to buy StorSimple, its almost never one incumbent storage vendor versus another competing vendor. It ends up being an incumbent storage vendor versus the

Page 91

March 14, 2012

Industry Report

cloud. And of course, the cloud providers you end up seeing are the kind of vendors like Amazon, Microsoft, AT&T, Rackspace, who end up being the vendors on the other side. The key point is that theres a business model disruption that is happening. You have these other larger companies who have learned to build and run these large-scale systems well at scale, as a service. And they want to go ahead and get into where the traditional vendors are operating. From an enterprise perspective, if Im a sales rep for a large incumbent storage vendor, I find it very challenging to go ahead and sell a device that is going to move all of the storage away from my portfolio into a third partys data center. So it becomes challenging for incumbent storage vendors to provide these cloud integrated storage solutions. Storage itself is a very competitive market among the big players because you dont have any one player who has a monopoly. So theyre all intensely focused on competing with each other, but at the same time, its a humongous storage customer spend. Enterprise customer spending on enterprise storage, when you combine the hardware, the associated software for the storage hardware, and the service and support, is more than $75 billion a year. That segment alone has more than $50 billion in gross margin. And among the major cloud providers, Google has most recently recognized that getting the data is important on two counts. First is, you have a tremendous incremental revenue opportunity, but the second is the whole battle of cloud computing is around the data. What iCloud is basically showing is once youre going to get your data in there, you are going to start using those applications in the compute and the devices around that. And thats the battle of cloud computing. When you look at what Amazon does to demonstrate cloud computing adoption, it doesnt talk about the number of VMs (virtual machines). It doesnt talk about number of customers. It doesnt reveal revenue. The only metric that Amazon will publish about how well that cloud is doing is the amount of data in the cloud. And as theyve been showing consistently, its been doubling approximately every nine months. And by many estimates, the close to 570 billion objects that they have now represents about 1.1 exabytes of production data thats already in there. What is happening amongst the key market trends is that the incumbent vendors are really focused on competing against each other. The enterprises are looking for an alternate way of dealing with their data. And you have these big cloud computing providers whove realized because of what has happened in the consumer side that the battle is about access to the data. If you have the data, then apps and compute will follow. And so you end up having a big set of players on the incumbent side. You end up having a big set of cloud providers, who are emerging. Theyre exceptionally supportive of companies like StorSimple, who are bringing the infrastructure to connect these clouds into enterprise data centers. TE: And have you tried to estimate the size of the market that youre trying to grow or develop? SS: The current version of StorSimples products, the ones that we started shipping in 2011 cover about 40% of the data in a classic enterprise data center and about 70 or 80% of the data for midsize enterprises. When we roll it up, our total addressable market is about $30 billion. And the way our products are sold, this market gets split between us and the cloud service provider. Because StorSimple products dont operate in isolation, they always require you to use a cloud storage service. StorSimple ends up getting about one-third to one-half of the customers spend over a three-year period. The rest of it basically goes to the cloud service provider. TE: And what are the key technology trends that you are leveraging within your solution? SS: Besides the transition around the cloud that we just mentioned, the other big trend that is a big enabler for StorSimple is the customer desire to collapse different types of storage systems that exist in that environment. There are some other companies which have already started doing that but theyve been doing it in a head-on, competitive model versus traditional storage vendors. In the case of StorSimple, well identify our workloads. And for those workloads, we will go and present to customers how we are able to combine primary backup, archival disaster recovery and tape elimination in a single package with a single device, not a pair of devices, a single device that seamlessly integrates into the cloud and then allows you as the customer to use the cloud for disaster recovery. If you have a disaster, the cloud itself becomes your backup data center for a disaster. If you have a primary production infrastructure, you can go ahead and, in our model, start using the cloud for staging, testing and development of most capacity for your existing data center.

Page 92

March 14, 2012

Industry Report

Besides the transition around cloud computing, we think that the other big trend is, at least for Tier 2, Tier 3 primary data, customers are more than willing to go ahead and consolidate the different sorts of storage that are used to manage data into a single appliance. TE: And what specifically would you call the secret sauce part of your solution? SS: If you look at what has happened in the past decade, Data Domain made de-duplication a common term, EqualLogic did iSCSI (internet SCSI), 3PAR did Thin provisioning, the similar contribution from an innovation perspective at StorSimple is a cloud snapshot. What a cloud snapshot does is that it fundamentally drives what I refer to as cloud data portability, which is data which is today trapped in your data center in an array of one vendor, and can only be replicated to another array off the same vendor. We really take the concept of that data or, very often what storage administrators will call a volume, and we take that volume and make it portable across data centers, across cloud providers. The result of that is if I have a disaster, and I may have hundreds of terabytes or petabytes of data offsite, the speed with which you can recover in the context of a cloud snapshot its several orders of magnitude better than what the traditional approaches offered in terms of recovery time when it came to tape, or in terms of cost compared to a traditional replication model for storage. TE: Your solution seems to be pretty unique. Who do you come into competition with? And, if you do come into competition with others, who are those? SS: Our primary competition ends up being the status quo. Typically, the way we uncover our opportunities really comes from either one or two scenarios. Its either one of the major cloud providers, who will take us into their customers. Or itll end up being resellers who are selling traditional storage solutions, who basically want to go ahead and sell a cloudenabled storage solution. When it comes to the competition, its almost always customers choosing to buy more of the traditional storage solution, or to opt for the cloud. Of all the deals that we have right now, theres not a single deal where the customer is considering another startup. Its pretty much almost do we buy more traditional storage? Or do we buy, more of a traditional storage, but also a little bit of StorSimple? If Im doing a $5 million purchase, should I just do all $5 million of my regular storage? Or should I do $4 million of regular storage and spend maybe half a million dollars on the cloud approach, and see how it works out for me, and see where else I can kind of go ahead and use it. In general, storage is a business where the vendor holding the support contract for a customer ends up getting the purchase order 90% of the time. Nobody gets fired for buying expensive storage. People get fired if they lose data. This is the reason why you have a market which is $50 billion in gross margin without a monopoly. You just have to look at what has happened on the servers and the virtualization side. The total server spend which would support software like VMs and all of that combined is under $25 billion. Gross margin in that segment is basically something like $7 or $8 billion, which is a fraction of what the gross margin in storage is. So the only way for new companies to go ahead and build big businesses is really to have a product that is rock solid, that works as advertised, that people can start out with small, experience it, and then, you have to let the product really drive customer loyalty and stickiness with repeat buying, or in the mid-range segment, its channels and referrals. With larger enterprises, its really repeat purchases. TE: OK. And, turning your customers, you seen traction in any specific verticals or customer geographies or anything that you can share with us? SS: Currently, most of our business is focused in North America. We have just started some expansion, and have customers in the U.K. and Japan. What we found very interesting has been that some of the top verticals for us very often end up being verticals which you would not necessarily associate with new technologies. And thats what was there in the beginning. Now its pretty horizontal, but normally people would think of financial services and technology as verticals. And what weve found is that some of our earliest Fortune 1000 customers ended up being manufacturing, oil and gas, engineering and construction services and media and entertainment.

Page 93

March 14, 2012

Industry Report

We also found a lot of traction with nonprofits and higher education. Most surprising has been the government sector. Weve had governments like the state of California, for example, go from a first meeting to buying and deploying production of our systems within 16 weeks. The CIO of this particular unit at the state of California, his point was that either I could continue paying my vendors and firing all of my employees, or we had to find a way to reduce costs and what you offered with the cloud. And in this case, they ended up using Microsofts Azure cloud with the StorSimple plans and ended up just completely changing the dynamic from an economics perspective for them. And they fast-tracked that out from a rollout perspective. So the point that Im trying to make is that its a pretty horizontal sort of offering. But at the same time, enterprises and organizations that really care about optimizing their infrastructure and having lower costs while driving agility end up being much more keen to buy our products than enterprises with workloads where they dont mind paying a million dollars to shave off an extra 10 milliseconds or 5 microseconds off an operation. The latter is not the kind of customers that buy us. Its really the former, who are looking to optimize infrastructure, drive agility and manage costs. TE: You talked about a few cloud service providers. Is that your primary go-to-market model? And can share with us who has been the typical partner when youre in the market over your brief history? SS: Yes. For the past year, most of our deals have happened with Amazon and with Microsoft. We did well enough at Microsoft that in 2011, Microsoft gave us a partner of the year award. That was simply based on the kind of customer traction that we saw in helping Microsoft get customers onto Azure. Now going forward, we see AT&T, we see Rackspace, we see Google, and weve seen HP and Dells services and businesses. IBMs done an OEM with another cloud service provider recently. Were definitely beginning to see a much greater set of cloud providers that are coming together and helping us out. The way we sell is an enterprise customer, a mid or a large enterprise customer, will buy the StorSimple appliance either directly from StorSimple or from the preferred reseller channel partner. Then they will go ahead and pick their choice of cloud providers. If theres a reseller involved, the reseller can pick a cloud provider, based on often the SLA (service level agreement) as well as the incentive that the cloud providers provide the resellers and package the solution or give it together. Or typically today, what we find is the end customer will pick one or two cloud providers to use with the StorSimple appliance and can upsell. So we dont end up having an OEM path as yet. Its much more of a joint route to market, though one could argue that, its along the lines of the iPhone/AT&T kind of a model, where if you look at what has happened with mobile phones or with, for that matter, Cisco routers, you can have service providers like AT&T move large volumes of product. But that happens when the market matures further. Right now, its much more along the lines of customers buying our products from us or our resellers, and then choosing their cloud provider services to use with our appliances. TE: Interesting. And if you look ahead, say a few years from now, what is your vision for the business? SS: Fundamentally, this is what I started out with trying to articulate using a lot of very technical terms in terms of what is that we bring for our customers. And now, its just come down to with what Apple did in 2011, its really building an iCloud for the enterprise. If I had to articulate what we do, we enable enterprises to build their iCloud. There are two components that go into that. One is about data accessibility and presentation, where how you connect all of your mobile devices to your existing storage systems and your existing applications. And we think there is a strong ecosystem of solutions coming to take care of that. And the second is how do you provide the cloud data management to drive the cost and agility of cloud computing and integrate the third party cloud services seamlessly into your data center. We end up basically delivering the lacking component to help enterprises build their iClouds. TE: I know youre a private company, but can you give us a sense of your size and historical growth rate? SS: So the companys a pretty young company. Weve gone through three rounds of financing. Our investors include Redpoint Ventures, Index Ventures, Mayfield, which was the first investor in 3PAR, and we have stuck with them all the way, along with Ignition Partners. So people definitely want to support the company for the long term. Weve been around for only two and a half years. Weve had a product in the market only for three quarters. Weve gone ahead and beaten all

Page 94

March 14, 2012

Industry Report

the initial sort of forecasts that we had and have really gotten into a very aggressive mode of hiring from a sales and marketing perspective only in the last couple of quarters. So its a fast growing company, but still pretty small with less than 100 people. Having said that, we go back and compare how we are tracking from a metrics perspective, with regard to both revenue at this stage of the company as well as amount of money spent by the company. The company is tracking at par or better than some of the big names in the business like EqualLogic and Data Domain. And thats something that I personally benchmark us by. TE: And when you look ahead, what do you see as your biggest challenge? SS: Our biggest challenge is also our biggest opportunity, which is that our solution is really tied to the hip with cloud computing and about how cloud computing impacts enterprise data centers and the storage infrastructure. The good is that, we enable enterprises for the cloud. Our challenge also comes from the fact that we have to work with the cloud providers to bring their solutions to market. A lot of these cloud providers end up being very large companies. From the time they say they want to go and do something, to the time they can find full alignment, with regards to sales marketing channels, along with incentives and compensation, it takes a while. One of the things that we want to be very careful and cautious about is to not race to market with partners before all the ducks are in the row. Our challenge really ends up being how do we go ahead and identify and work with the right partners in the ecosystem without trashing ourselves and our customers and losing focus. And this is where its been very rewarding to see how what Amazon and Windows Azure from Microsoft has brought in the enterprise focus in 2011, both in terms of the people theyre hiring, in terms of the kind of sales and channel incentives that theyre putting in place, that has made it very rewarding to go ahead and really make a lot of these enterprise deals happen. Our average sales cycle is 16 weeks. When you overlay that with the fact its mainly into the enterprise, it ends up being very interesting. The fact that it is an early stage company, people have to do a proof of concept of a product before they will buy. If it wasnt for the kind of support that we get from cloud providers, it would be very hard for us to go ahead and get dollars coming to us, which were originally destined for traditional storage vendors. TE: So this has been a great overview of StorSimple. Is there anything else you would like institutional investor clients to know about StorSimple? SS: I think the only other thing that I would add is that we find ourselves to be in a very unique position because StorSimple as a company started out right at the bottom of the recession. A lot of people remember it as the great recession of 2008. We formed in the spring of 09. General Motors had just become Government Motors. We find ourselves to be in a very unique situation where we dont have any other startup competitor which is bringing the same proposition to market. Just to put that into perspective, when we got started, storage was not an attractive investment neighborhood. All of the public companies were available for great prices and were trading at par or often below their IPO prices like 3PAR; EMC had not acquired Data Domain when we got started. And then, you did have 3PAR and Isilon acquisitions and Fusion-ios IPO. Clearly, the investor perspective and thesis has changed. You can see how you have such a tremendous amount of investor momentum behind essentially everything around SSD. But if you take a step back and see what the big trends that are happening are one of the biggest trends that are happening is around cloud storage. You find that, there is one trend around scale-out storage and SSDs and scale-out storage from companies like Coraid. There are solid state storage, which is where companies there are companies like Pure Storage, the Violin Memory and Nimbus. Then you end up looking at enterprise cloud storage as a third big trend. We find ourselves to be very uniquely positioned as being one of the only companies which delivers our proposition, and one of the very few companies in that segment. Its a great position to be in, where you have incumbents which seem to be in the midst of a business model transition to position themselves with how they work with cloud computing. And you have the cloud providers, who are very eager to partner with emerging infrastructure companies like us. And not having too many other startup competitors has really created a very unique position and is getting us a lot of momentum. That and the fact that theres just tremendous ecosystem support. When we go into an account, its never an EMC or a NetApp account manager versus a StorSimple account manager. It is almost always an EMC account manager

Page 95

March 14, 2012

Industry Report

versus a cloud provider account manager, whos trying to go ahead and tell the CIO how he can help optimize that CIOs infrastructure and make it more agile by bringing in a cloud service. What StorSimple does is make that whole transition seamless. So right now it happens to be the only cloud storage appliance that is certified to be primary storage by both Microsoft and VMware. Weve had this position for a whole year, and no one else has come close.

Page 96

March 14, 2012

Industry Report

Dan Scheel, President of Texas Memory Systems Texas Memory Systems, Inc. designs and manufactures enterprise data storage and digital signal processing systems. It offers rack mount flash storage, rack mount RAM storage, peripheral component interconnect bus storage, digital signal processing, interfaces, and legacy products. The companys key markets include the financial, department of defense and intelligence community, high performance computing, telecom, ecommerce, and healthcare industries. Dan Scheel is the President of Texas Memory Systems (TMS). Dan joined TMS in 1997, rising through the ranks to become Vice President of Engineering in 2009 and then President in 2010. In addition to directing TMS daily operations, Dan oversees the company sales efforts and is directly involved in RamSan engineering. Here are excerpts of his interview with ThinkEquity LLC: Rajesh Ghai (ThinkEquity): Can you explain the key problem that youre trying to solve in the data center at Texas Memory today? Dan Scheel (Texas Memory Systems): Sure, the key problem that TMS is trying to solve in the data center is bottlenecks that occur from storage devices not being able to give their data back to processors fast enough and that is typically found in a spinning disk drive not being able to handle a large number of transactions per second or I/O. Or not being able to give a large amount of bandwidth of data back to the processor so that the processors can obviously do what they were supposed to do with the data. TE: Why do you think existing solutions from the larger more established vendors cannot address the problem that youre trying to solve? TMS: Well, I think that they can address the problem unfortunately though the solutions to get the same level of performance would require an extremely large footprint and be extremely expensive. A data point that I can give you is, in May of 2011 we were running storage performance council (SPC1) results and we were able to get a level of performance at about 5% of the cost of the nearest tested solution. So, it is possible for the larger, more established vendors to provide solutions. Unfortunately for the large vendors, its much more cost effective to go with our solution and the footprint that we provide is extremely small. TE: Thats wonderful and what are the key market trends that you see that are influencing demand for your solutions? TMS: Recently weve seen adoption rates of hard drive to SSD (solid state drive) increase and there are several reasons. Most recently, the floods in Thailand became a problem for hard drive manufacturers to be able to supply their product. That has been driving some of the SSD adoption rate, but in addition to that, there are other things that have been there for months and years; which are better performance numbers. When you are comparing the dollars I spend for a diskbased system versus the dollars I spend for a solid state system, you get more performance out of an SSD. As well as things like power consumption the amount of power it takes to provide a system that is made out of flash is lower than the amount of power that it takes to provide a system that is made out of spinning drive. And then also the footprint, the density that you can get with an integrated circuit, the flash chips, versus spinning drives. TE: Great and have you put any thought into the size of your addressable market at this point of time? TMS: Well, it is kind of interesting. Obviously its very much forward-looking and there are a lot of different analysts that have taken a stab at what it could be. I think some of the numbers that Ive seen seem very, very realistic to me with the total amount of enterprise solid state disks being somewhere around $1 billion right now and going to about $5 billion maybe by 2015 and that seems very realistic. As well as when you look at the entire solid state disk market all the way down to the consumer grade and all the way up to the enterprise being something that could go to say $10 billion or $12 billion by 2015. So theres obviously a lot of potential in the market and TMS is trying to address that with our technology in the enterprise space. TE: Can you talk about the key technologies that you are leveraging in terms of building a solution? TMS: Sure. Normally the key technology that we get judged on is a legacy thing out of the disk drive-based system. So simply looking at how much can I spend? What are the dollars per gigabyte that I can get out of a solution? We are typically a dollars per performance type of solution.

Page 97

March 14, 2012

Industry Report

When we look at technology that utilizes our core performance its normally some form of structured data and that typically translates to be some form of a database whether its an Oracle database or a SQL (structured query language) database, something where your data is very much structured and you can therefore get a lot of return on investment if you are using a high-performance storage device to run that application. TE: OK, there are quite a few other vendors out there in the market that are leveraging flash and what I really wanted to understand was what your key differentiation is and what your secret sauce is? TMS: I would say the secret sauce that you can identify what Texas Memory Systems does well is we are building it from the lowest level up. And what I mean by that is you can look at other competitive solutions that are using hard drive form factor SSDs and they are putting it inside a RAID enclosure that was used for disk-based systems or spinning disk systems but they are now just trying to increase the performance. So what that means they are using old building blocks to try to start to have better and better performance. And that will give you some amount of performance but youll immediately hit bottlenecks and youll immediately hit something else that you have to overcome to get better and better performance. TMSs secret sauce is the ability to take that flash chip at the lowest level and say, Im going to balance the amount of flash that I have as well as the interfaces that access that flash so that as application servers are requesting data they can come into our flash appliance and immediately have access to all of that bandwidth or all of the number of I/Os per second that they can get. Its really a designed solution around the flash rather than using Intel processors in a box or basically having to put together other pieces, other commercial off the shelf types of hardware in order to get a solution. TE: And who do you see as your nearest competitors and how do you see competition evolving over the next few years? TMS: We do compete against the spinning disk drive types of companies so the EMCs, the NetApps. Probably the closest competitor that makes a flash-based system similar to us would be someone like Violin, but where we see our system get installed is obviously we tend to replace spinning disk drives quite a bit so the EMCs and the NetApps are someone that we would be replacing with our solution. And in terms of running into Violin, we do see them every once in a while. What we tend to find out is that our buyers or customers are very smart consumers that typically like to do, we call them bake offs, where you bring the competition inhouse and you perform tests on your workload on the box to see how it performs. And we do quite well on those types of situations. TE: Great. You talked a little about your customers; can you tell us who your target customers are, any specific segments or verticals where you are particularly strong? TMS: Sure. We target anybody that needs extremely high performance. We are strong in Oracle database type environments and specifically financial markets. We also do quite well in e-commerce, high performance computing, as well as telecommunications. Its kind of interesting because our technology is not necessarily one vertical, but the high performance of many verticals and that has worked quite well for us because at the end of the day what it boils down to is were trying to accelerate an application for a company. Normally that application is a revenue-generating application extremely critical to the company to both go as fast as possible to generate money for them but also reliable as possible. So our customers who we service are looking for solutions to basically push that envelope and make it to where they can make more money off a storage device accelerating their application. TE: Can you share with us a couple of your key customer wins and how you went about winning those deals? TMS: Obviously we have a large amount of online success stories but I will touch on just a few. The main reason I think that we are successful in how we go about winning the business is letting the customer understand what their data workload looks like. And what I mean by that is, obviously your application is reading and writing data but you dont exactly know how its reading and writing data and how you can improve the performance of it reading and writing that data so you can obviously get more performance out of it or go faster. An example that I have for you is theres an online gaming community that uses a game called EVE Online. And this is a type of game that is played 24/7. You basically have users that pay a monthly subscription and they login and they play

Page 98

March 14, 2012

Industry Report

the game, when their done playing they log out and the game continues to play. They came to us because they were having a problem with the concurrent users that they had logging into the system. I dont remember the exact number but it was maybe somewhere around 20,000 concurrent users logging in and playing the game. Now if you think about it, playing that game is just a structured database of queries going on in the background. I believe the game is kind of a spaceship-type game so how many shields does my spaceship have is just a database query or what is the properties of this thing that Im investigating, thats just a database query. So we were able to come in and provide them a way to speed up their database queries and therefore improve the performance and improve the users expectations of what was going on. What I mean by that is they would hit about 20,000 concurrent users, users would login, and the game would slow down, it would lag. And the reason it was lagging is those database queries were slow to respond and therefore people were getting upset that, well I dont want to pay for a game that is slow. I want a game to be responsive and not have to be sitting around waiting for it. So we were able to come in and provide them a high-performance solution that allowed their database queries to be in the hundreds of thousands. What that meant for them is they could increase the amount of concurrent users. Since this is a game-based subscription/monthly subscription type of a revenue-generating application, it allowed them to have more and more customers that could login to the game and play it simultaneously, which ultimately meant more revenue for them. So it is a very easy, kind of return on investment type of calculation to show if you get this many more concurrent users, you obviously are making this much more per month and therefore paying for the box extremely quickly. TE: Thats very interesting and you talked about customers and your go-to-market model. How do you reach your customers and if you can talk about your key technology and go-to-market partners, that will be great. TMS: OK, so what we have is what I would call a channel-assist program in the fact that leads come to us and we obviously feed those to the channel but there is a lot of education that goes on from our staff to assist them in figuring out what is their environment look like right now. I talked earlier about that workload to your storage device and helping the customer or the end user figure that out so they can understand whether it makes since to be in an SSD or not or specifically in a TMS RamSan SSD or not. And so our channel-assist go-to-market model makes it to where we are actively helping the channel educate users on how to use it, whats different about it versus other people or other competition and that has been very successful for us in terms of being able to get leads that come to us as well as educate people about our technology. Concerning the key technology and the kind of go-to-market partners that we have, were certified behind IBM storage heads, like the SAN (storage area network) Volume controller. We obviously are used quite a bit in Oracle or say behind BlueArc-type of storage heads. There are key partners that utilize us as that high performance storage when the customers cant get the performance out of their disk based solution. What that means is typically we find that customers are coming to us because they have a problem that they need to solve and its normally a performance issue-related problem and were able to provide them a solution that is extremely cost-effective with the amount of performance that we give them. TE: Thats wonderful, and if you look ahead, look towards the future what is the vision for the business say a few years from now? TMS: So we still want to continue our our performance leader kind of dominance that we have of being able to say we are the worlds fastest storage that we can provide solutions that are extremely cost-effective when you compare to what you can do with other technology. But I think looking forward of how we continue to grow Id like to see us offer solutions that are designed for all the different levels of the business so, obviously right now being extremely high-performance we have our customer base being enterprise-type customers. But theres obviously a lot of potential or possibility for the smaller businesses that are not enterprise but still business class type of customers. TE: And from a product roadmap perspective, can you talk about what innovations youre going to bring to market in the future? TMS: So one of the things that we just announced this past Tuesday was a new product - its our RamSan 720 and it is the first single system, no single point of failure box for us. And the reason that we think that that is key is there are typically two schools of thought on no single point of failure.

Page 99

March 14, 2012

Industry Report

Theres the school of thought of you cannot have that in one device. You can obviously design for it, but you cant have it in one single system. And the reason I point to that is we see that people install mirrored installations so that they can say, if something happens to one side I at least have my other side up and running and therefore have no single point of failure. But there is also a class of enterprise customers that do have the comfort level of saying that having a box that is designed for no single point of failure is acceptable and I do understand that if I have a, say an issue inside the box that I might have reduced performance but I obviously do not lose access to my data. So thats something that we just announced. Our previous boxes were designed with the intent of if you wanted a no single point of failure solution you would install them in mirrored pairs. In addition to what we just announced, I think youll see from TMS in the future more features that are surrounding data management in flash. And what I mean by that is disk drive-based solutions have been around for quite some time and there has been a lot of software written to optimize what goes on with disk based data and moving and managing that data. And I think well start to see more of that coming from TMS, but more on the flash-based systems of moving and managing flash-based storage systems. TE: And can you give us a sense of your size and your revenue projections at this point in time? TMS: I will not be able to answer the revenue projections and things like that because that exposes, obviously what weve done in the past and being a private company we dont provide that. But what I can tell you is the size of the company. Were about 100 employees. The majority of the employees are engineers; about 40% of the company is engineering which allows us to basically take our double digit growth that weve seen over the last several years in flash. To take that type of growth profit and reinvest it in R&D so that we are continuing to push the envelope, continuing to be that high performance leader and have the worlds fastest storage. TE: Great and as you look out the next couple of years what do you see as your biggest potential challenge? TMS: Id probably have to say educating potential customers, educating the people that are looking at the marketing of competition or just simply the marketing of the hard drive vendors as well as the flash-based vendors and educating them so that they understand why different storage that is made out of flash is actually different and its not going to behave the same and understanding that theres orders of magnitude types of difference that you have the ability to say if Im flash vendor A, Im definitely faster than the fastest disk-based system. Thats actually quite easy to do if youre using flash and its quite easy to be faster than spinning disk. But to explain what occurs with flash and why some solutions are different or better for the customer in their environment, what their data workloads are actually looking like, that I think is going to be the challenge of making sure that customers dont just naturally assume I know it all, its like disk its just simply a chip instead of a disk and educating them on what is actually going on inside systems. TE: Thats great, Dan, this has been an excellent overview of TMS. To wrap up, is there anything else you would like to let our institutional investor clients to know about TMS? TMS: Sure, TMS has been around for 33 years and the entire time we have been doing solid state disks. We started out with RAM (random access memory)-based solid state disk in our shared access memory systems and weve come all the way through those 33 years and now the majority of what we sell is a flash-based solid state disk. In that time theres been a lot of development, a lot of things that were learned about how you create an extremely high performance, extremely fast bus architecture that allows the application server to actually see the large level of performance that we can get. In addition to that, what I think Id like people to know about Texas Memory Systems is we are a company of engineers; we look at things in terms of problems that need to be solved. Whether that is what a chip manufacturer give us as a commercially-available chip and how we use that chip. We are very good at designing solutions and designing products and therefore that means that we are very versatile in what we can do, what we can design, that its not just simply build a product and try to sell that product for 10 years. We are constantly innovating and constantly pushing the envelope and obviously receiving patents for what we do. TE: Great thank you, Dan, and I wish you all the best and look forward to tracking you in the future.

Page 100

March 14, 2012

Industry Report

Kieran Harty, Co-Founder and CEO of Tintri Tintri, Inc. develops purpose-built storage systems for virtual machines. The company offers Tintri VMstore, a virtual machine aware storage appliance that enables enterprises to manage and store virtual machines. It also provides purpose-built solid state drives; and Serial ATA hybrid storage arrays. The company serves various companies in electronics, mobile games, geo-informatics, technology, and higher education industries. Mr. Harty was previously Executive Vice President of R&D at VMware for seven years, where he was responsible for all products. He led the delivery of the first and subsequent releases of ESX Server, Virtual Center and VMware's desktop products. Before VMware, he was Vice President of R&D at Visigenic/Borland and Chief Scientist at TIBCO. Kieran has a Ph.D. in Electrical Engineering from Stanford University and a Master's Degree in Computer Science from Trinity College Dublin. Here are excerpts from his interview with ThinkEquity LLC: Rajesh Ghai (ThinkEquity): Kieran, can you explain the key problem you are trying to solve in the data center at Tintri? Kieran Harty (TT): So, I am sure you are aware theres been a huge trend towards virtualization in the data center. And now its the case that there are more workloads being virtualized than are running in physical environment. So customers have got a lot of benefits from virtualization on the server side. Theyve reduced the number of servers that they are using by a factor of ten or more and got a lot of benefits in terms of agility. But they havent seen the same kind of benefits on the storage side. So storage is expensive. And typically 60% of the cost of a virtualization project goes to storage. Its complex. Its expensive to manage. And its often an impediment to deploying applications. And storage is often something that prevents applications from being virtualized for performance reasons. Databases may not be virtualized in enterprises because of performance concerns. The three problems that we are attempting to solve are the complexity, the cost, and the performance issues associated with traditional storage. TE: Great. And talking of traditional storage, why dont existing solutions from existing legacy vendors address these problems as well? TT: There are two issues behind this. The first is that the market is dominated by EMC and NetApp. Those systems were typically designed 20 years ago and they were designed for the physical environment and not for virtual environments. They were also designed for disk-based systems and now, from a technology perspective, we can take advantage of things like SSDs (solid-state drives) which radically changes the way that you can build storage. The combination of having being built for the physical world and being built for disk-based technology mean that the problems cant be solved in the same way. TE: Great. And you talked about storage for virtual environments. What is the size of the addressable market that you are targeting with your solutions? TT: That market today is $11 billion and thats an IDC estimate. That market was zero in 2001, and is growing with virtualization. If you look at the storage markets, there are few areas that are growing. Virtualization is the fastest growing, and big data is also another area where there is growth. TE: Besides solid state drives, what are the key technology trends that you are leveraging in your solutions? TT: Obviously virtualization is the huge one and the trend towards virtualizing all workloads. From a hardware technology perspective, its 10 Gigabit Ethernet. Historically Gigabit Ethernet was too slow, and most performance-critical workloads would be deployed on Fibre Channel. Then on the other side, multi-core processors allow the building of systems using off-the-shelf hardware to do things like inline deduplication and compression that are very compute intensive. So its a combination of the three; the SSDs, the 10 Gigabit Ethernet and multi-core that allows you to build very different solutions. TE: Can you provide some more detail as to how your solution addresses the problem that you referenced earlier?

Page 101

March 14, 2012

Industry Report

TT: Sure. If you look at traditional storage, traditional storage is not built for a virtual environment, and in those environments, with traditional storage, you tend to have to manage things like LUNs and volumes and zoning and a lot of arcane technology concepts that are very much associated with the physical world. We swept that out of the way and say that what you should really be managing is virtual machines. So the first thing weve done is that weve built a storage system that is specifically for the virtual environment. And also, as part of the same storage system, its a storage system thats built for Flash SSDs. So its a combination of two things. Its a follow-up system or storage system that is built exclusively for virtual machines, and its also one thats built for Flash. The combination of those two allows us to solve the problems that I described earlier of complexity. Our solution is simple performance. There is a lot of performance associated with what we do. And its also a form factor issue and less expensive than existing storage solutions. TE: Great. And you talked about storage for virtual environments. Are there any specific workloads where you are seeing early traction for your solutions? TT: There are actually three areas where weve seen traction. The first is something called VDI, or virtual desktop infrastructure, which is essentially about moving desktop workloads into a virtual environment. In the past, thats been an area where storage has been a very significant impediment to deployment in terms of cost and complexity and we solve that. We have a number of customers including people like Washington State University, also a company called SemmesMurphey which is a clinic, which are using our solution for VDI. The other one is private cloud environments where people or companies want to have an internal cloud, where that they can run a variety of different workloads. We have a customer called Digital Chocolate, who is in the gaming space for the iPhone and for Facebook. They are running a private cloud thats using a Tintri system. And they have over half a million users that are playing games on their private cloud every day. The third one is databases where customers have an issue with performance of databases. One of our customers is TIBCO and they had problems with running Oracle Financials and also a SAP analytics application in a virtualized mode. And they were easily able to solve those problems by running on Tintri. TE: Who do you see as nearest competitors and, more broadly, how do you see competition evolving given that you see a lot of startups using Flash and targeting the virtual and storage environment? TT: I think the major competitors are definitely the incumbents; so predominantly EMC and NetApp. We also see a bit of HP and Compellent. Then in terms of the Flash base, we are actually the first product in the market today that has enterprise features including things like high availability and snapshots that actually is an array that uses design from the ground up with Flash. We would expect that there will be additional players that go after the Flash markets but they appear primarily focused on general purpose storage. Our message is exclusively about being specialized for virtual environments. We are not aware of any array-based systems that actually go exclusively after the virtual environment. So expect that competition will tend to go after the performance end of the market. And as we are successful, we may see people trying to move in this direction but so far that hasnt been the case. TE: Great. You briefly touched upon some of your customer wins. Can you talk a little bit more about what those customers were looking for and how you went about winning those deals versus the competition? TT: The major issue and this is of course very significant for a startup is there has to be pain. If we consider something like Digital Chocolate, which I talked about earlier, the pain that they were experiencing was they had been using Amazons EC2 to run their games. They decided to bring things in-house and they were running on a SAN storage array. The pain that they were experiencing was one of complexity and it was difficult to manage. They were also experiencing performance issues, which were difficult to diagnose, and it was also becoming expensive to manage things. So the way we approached them was by talking to them, doing a demo for them initially, demonstrating the simplicity of the product and then showing that it could perform. One of the things that was particularly compelling in this case is it is

Page 102

March 14, 2012

Industry Report

very easy to move on to our technology. They moved on to our technology literally in an afternoon and were running in production extremely quickly. So its primarily about pain. Another one would be the VDI case where Semmes-Murphey wanted to run VDI and there wasnt a storage solution in the market that would actually meet their needs at reasonable cost. They were easily able to run on us. Again, theres a lot of pain out there associated with storage for virtual environments. Doing a demonstration of what we do causes the lights to go on and then we can do a very quick proof of concept with the customer. TE: Thats wonderful. And whats your go-to-market model as a startup? TT: Sure. Our go-to-market model is that we have a small direct sales force but we work through the channel. Our goal is to have almost all of our deals if we can, go through the channel through reseller partners. The majority of our deals do have a reseller involved. We are expanding our reseller network in Europe and we are already in Europe, we are exclusively working through distributors and resellers. And in Japan, we are working closely with a reseller. So it is very much a reseller and channel-driven model. TE: Can you talk about your key technology partners? TT: We are using off-the-shelf hardware. We are currently using Intel motherboards and we are using Intel SSDs, as well as Intel 10-gigabits NICs, but we are also designed to be able to take advantage of whatever else may come into the markets. TE: And if you look ahead, what is the vision for the business say five years from now? TT: As I mentioned earlier, the market is very substantial at $11 billion. This is the greatest level of technology change that has occurred in the storage market for decades or for 40 or more years. We think that there is an opportunity to build a large independent company. Our vision is about enabling virtualization and enabling people to build private clouds, and we think that we can have a very large independent company which is based on that huge market, and the pain that exists there. TE: Can you give us a sense of your size and the historical growth rate in that market? TT: We can say that weve grown as much or we have grown faster than the storage vendors that are in the same space. We reported that we grew 100% quarter-over-quarter. We have over 50 customers who are in production now. TE: When you look ahead over the next couple of years, what do you see as the biggest potential challenge? TT: The biggest potential challenge is that the Flash market is very confusing. Our belief is that there will be a fragmentation of the storage markets. Historically, the storage vendors were pushing a single solution that would work across a variety of different applications. There appears to be a fragmentation that is occurring there, whether its a solution that may be appropriate for high-end performance, big data and for virtualized workloads. We think that theres quite a lot of education that would be required there. And its probably confusion, and just the fact that there is more technology change than I think customers have been accustomed to. Thats probably our biggest challenge. TE: Thats good to hear, in a way. But when you look at the entire market, what do you see as the future growth of this market? Its a nascent market but how do you see this market evolving over the next few years? TT: Well, its going to evolve with the deployment of virtualization. So again, if you look at it, virtualization is growing at about 25-30% a year. As virtualization grows, its absorbing or its creating new ways to do things. Backup is different in a virtualized environment and disaster recovery is different in a virtualized environment. The market for VDI is very, very nascent and there are hundreds of millions of desktops out there that are not virtualized today. We think that with the combination of growth in the existing markets, the use of VDI and new ways of providing disaster recovery, that there is enormous growth potential in this market.

Page 103

March 14, 2012

Industry Report

TE: Thats a good overview of Tintri. As a wrap up, is there anything else you would like our institutional investor clients to know about Tintri? TT: One thing I would emphasize is that building storage is hard and its not something that can be done quickly. We have a very talented team. Our VP of engineering came from Data Domain. Our engineering team comes from VMware from NetApp, Data Domain, and Brocade and building out this kind of technology is difficult. And meeting the liability requirements is also difficult. So there are substantial barriers to entry for vendors entering this market and if you started doing something right now, you are basically too late. TE: Thanks Kieran, thanks for your time. I really appreciate it. Good luck and I look forward to tracking you in the future.

Page 104

March 14, 2012

Industry Report

Daniel Crain, CEO of WhipTail WhipTail is an independent manufacturer of solid-state storage arrays designed to drastically improve the scalability and performance of applications such as virtualization, databases, financial trading and email. WhipTail is currently a leader in extremely high performance, pure solid-state storage arrays for the commercial, enterprise and government markets. WhipTail's storage arrays are installed worldwide, in numerous mission critical virtualization and database environments. Its datacenter and virtual desktop systems dramatically improves write performance, reduce delays related to disk contention and access times thereby allowing database or virtualization servers to process more data in significantly less time, while lowering the overall TCO by up to 90% when compared to traditional storage. Dan was an original founding member of the StorageApps team, where as Executive Vice President, he built and ran the engineering, manufacturing and services organizations, that created the first Storage Virtualization product. He was most recently at Brocade Communications, where he was Chief Technology Officer for many years. Here are excerpts of his interview with ThinkEquity LLC: Rajesh Ghai (ThinkEquity): Can you explain the key problem youre trying to solve in the data center at WhipTail? Dan Crain (WhipTail): Sure, the principal problem we are working on right now is the next generation of storage and the performance characteristics that are required of it. For probably about 10 years there has not been a huge improvement in the behavior of the commercial data storage systems. And only in the past couple of years has the performance gap really become substantive enough, as an industry, where we really need to do some work in the areas. So our principle value proposition is we are using NAND flash technologies in a traditional storage form factor to radically raise the performance envelope for commercial and enterprise storage. TE: Great. And why cant existing solutions from the larger more established vendors address this problem? WT: There are a variety of reasons. One of it is that the traditional vendors for really 25 years or more have been principally working on traditional magnetic mechanical hard disks which have problems in terms of meeting performance requirements. We havent had a lot of development, from a performance improvement standpoint, in the magnetic mechanical arena in about a decade when we last had a major speed improvement to the way they performed. And these are materially different than the hard disk that appeared inside of PCs and laptops. The enterprise hard disks have pretty much reached their limits in terms of performance. The traditional vendors are, for the time being and probably for several years in the future, substantially limited in their performance by their leveraging of hard disk. TE: You talked about some of the key trends within technology but what are the key market trends that you see influencing demand in your target market? WT: There are a variety of them and we see this every day in our sales and customer interactions. Over roughly the past decade, the total performance capability of processors, servers, networks and memory has radically improved by many orders of magnitude. And data storage systems have stayed pretty much flat, in regard to performance capability, or not that much faster than they were 10 years ago. As we have seen internet scale begin to creep into the way everything works in the industry, weve seen the lack of commercial storage and data storage systems ability to keep up with the demands for internet scale and very, very rapid performance is a driving trend in many areas. Simple examples are virtualization environments where one computer or one server, used to host one operating system in one application. Now that same server can host 50. That puts a tremendous demand on storage performance and in the overall behavior of the way storage and I/O (input/output) subsystems work. An example would be a customer database, when you log on to our website and you want to see your information quickly. The sheer volume of information being presented out to you is more and more every day and the demand on getting that presented quickly has become even greater than ever before. Internet scale is so substantial and the leveraging of large scale databases and virtualized environments has become so huge and critical that storage systems need to keep up. TE: Looking at your solution, what do you think is the addressable size of the market and how do you think the market will grow in the future? WT: Yes, thats actually a very good question. The available market is large and it doesnt represent the entire storage industry. People store data for a variety of reasons. Obviously they are storing it because they want to keep it but performance is not a requirement for all data storage. In many cases, cost effective capacity is a requirement for data

Page 105

March 14, 2012

Industry Report

storage. Our market is a growing market and the demand for high performance storage is a growing market. Its very difficult to talk about the actual addressable market. Its in the billions of dollars even if you look just at the VDI (virtual desktop infrastructure) as an example. The available market for products that we do, according to Gartner or IDC types of third party vendors, probably represents about $5 billion of the addressable market over the next few years. That doesnt include databases and that doesnt include web data systems. So the total addressable market is substantial and wide encompassing and its only going to grow over time. TE: And how would you define specifically WhipTails secret sauce? WT: Our systems were one of the first to be purpose-built to behave like a 100% solid-state storage array based around traditional, well understood NAND flash technology. Its important to note that NAND flash was developed in the mid1980s by Toshiba. It was really deployed in cameras and cell phones and then inexpensive USB flash drives that you get in the office supplies store. A number of applications that are totally different than the duty cycles required for commercial storage and NAND was never designed for this duty cycles. Our intellectual property does many things. It enables NAND flash to behave in a way that works with the duty cycles of commercial storage as it is very high performance. We do radical acceleration of data writing performance. We extend the life of NAND flash by many, many years, getting it pretty much on par with magnetic mechanical. We do a whole bunch of things to protect data integrity. NAND flash behaves very differently than magnetic mechanical disks do. So we have a very large amount intellectual property and a number of patents applied for in this space that are completely organized around the creation of systems that are large scale storage systems with NAND flash in it. TE: And who do you see as your nearest competitor and typically when you compete on a deal, do you replace the solution from one of your competitors or is it typically for a completely new use case? WT: Its usually for new use cases. Very often we partner with a lot of the larger traditional competitors, EMC, NetApp, those types of vendors. Their products are not typically built around NAND flash as a general purpose storage solution to replace magnetic mechanical, especially from a capacity cost standpoint. The market is just not there yet and there are really no magical technologies that will get the price point to the magnetic mechanical price point. So we often partner with the traditional storage array vendors. In terms of direct competition, our direct competition, more often than not is a Violin which is a very capable competitor as well. Occasionally we see systems sold by Hewlett Packard, Oracle or other major OEMs that have Fusion-io cards in it. But typically speaking, those are the only real direct competitors that we see. TE: And is there any specific customer vertical or segment or use case where youve seen initial traction? WT: Were doing very well in a variety of different types of database applications. Were doing very well with virtual desktops. We are, for all intents and purposes, the standard bearer in virtual desktops. We have wonderful partnerships with VMware, with Citrix, with Microsoft in this area. And were starting to be brought into some very interesting use cases such as high frequency trading environments where hedge funds have operations which are very much based on low latency market data tech measurements. We do very, very well here because our latency is not that much greater than traditional DRAM-type systems. So there are a variety of different things that were being introduced to on a weekly basis by creating the technology guide on the IT side of the business where we are looking for new things to do. TE: Can you provide us with a couple of your key customer wins and how you went about winning those? WT: A recent win was at Advanced Micro Devices. AMD uses us for our extremely fast cycle times to do firmware testing against raw CPUs. CPUs these days are extraordinarily fast and extraordinarily powerful and we are able to actually test them in final QA (quality assurance) cycles or when they get back first silicon from your fabs. They do a tremendous amount of testing on them and they need storage that can keep up and were really the only product in the industry that can keep up with their new model CPUs. TE: What is your typical go to market model? WT: We are traditional in that sense. We do a lot of direct demand generation. We have a very broad channel that includes over 110 partners signed worldwide.

Page 106

March 14, 2012

Industry Report

We distribute both in the U.S. and in Europe and we have a pretty nice channel in both locations and thats only growing. Our inbound channel additions have been accelerating as of late because we had a lot of a pretty visible wins in them with channel partners. And of course we have some of our sales folks who do spend time out speaking to larger customers directly. TE: OK. And can you talk about your technology and your go to market partners? WT: We are unique in that we are Citrix and VMware certified across the whole suite of our products. We have a lot of channel partners. We have a very nice relationship with a lot of the major storage OEMs (original equipment manufacturers) in the field and youll see some really, really unique and key new sort of go to market partners announced in the next 90 days from us. TE: And what is the vision for the business say five years from now? WT: I think that we will most likely be in a position where we are shipping some of the fastest and most reliable data storage equipment ever built. We are very, very resolute on being the engineering leader, on being sober about what we do, on being extremely ethical on business practices. It means everything to us because our customer base right now is nearly 100% referenceable and we intend to keep it that way which is a challenge as you grow. I think in the next several years were going to see new types of memory come out that are going to offer interesting alternatives to the NAND flash technologies. Were going to be on the cutting edge of that. And were going to sort of build our technology into the high performance data storage market that beginning to develop right before our eyes. TE: And can you, from a product road map prospective, tell us what we can expect from you in the near future? WT: I would rather not go into specifics but I will say is that youll see some things from us this year that will set new standards reliability, in performance, in scalability and really in a new way of thinking about the deployment of data storage subsystems in commercial and enterprise accounts that are the most demanding accounts in the world. Without saying too much I would say that we, from a product road map standpoint, have some things that will change peoples beliefs about how a storage system can be built. TE: And can you give us a sense of your size and your financial goals for the future? WT: We have an extremely solid balance sheet. We are nearly profitable from sales. Were obviously not from operations now because we are investing to build for the future. We have just closed a very nice series B round with some wonderful investors and there has been a decent amount of press on that. We are investing heavily in post-sales support, in customer service and we invest deeply in engineering. We are hiring engineers at a very rapid clip both in the U.S. and in India. In both locations we have excellent teams who work very well together and were a very solid company to do business with. TE: And as you look out over the next couple of years, what do you see as the biggest challenge? WT: I think there are a couple of things that I would think about. One of them is keeping up with the appetite for bigger, better, faster or cheaper. Thats where the challenge is in the commercial technology industry and thats never going to go away. We are a cutting edge company with a deep engineering culture and we are always working on evolving our sort of art form forward. And I think thats probably the single biggest thing that we think about every day. Its how are we going to always push forward the art form and push forward the edge of the envelope. TE: Great. And to wrap up, is there anything else you want to tell our institutional investor clients? WT: I think that every one of them should own our technology. We are widely used in trading and financial environments and we have an extremely high level of their visibility and I tend to think were unique and that we are extraordinarily focused as a company. Sometimes were criticized for being not a PR rich company but we are a very, very focused on customer satisfaction and doing the right thing. Part of my background is that Im a veteran of Wall Street IT at Morgan Stanley, Deutsche Bank, Scudder, and Prudential. So I know what it means to actually absorb these kinds of technologies and I know of the expectations. We are very focused around making sure we do the right thing. TE: On that note Dan, thank you for your time. We appreciate you taking the time to talk to us and that we look forward to tracking you in the future.

Page 107

March 14, 2012

Industry Report

Appendices

Note: In the following section, the company logos were sourced from the respective companies websites; funding data was sourced from S&P CapitalIQ.

Page 108

March 14, 2012

Industry Report

Appendix I: Private Company Profiles

Name: Avere Systems Headquarters: Pittsburgh, PA CEO: Ron Bianchini Phone: 412-635-7170 Year Founded: 2008 Website: averesystems.com Description: Avere Systems provides Demand-Driven Storage solutions that dynamically organize data in response to business demand. Its long-term vision is a storage network in which the delivery of data is no longer compromised by residing within the same storage tier as data retention functionality, both within the data center and out into the cloud. Investors: Menlo Ventures, Norwest Venture Partners, Tenaya Capital Financing History: (US$ in Millions) Date 9/2009 Amount Raised $15 Investors Menlo Ventures Norwest Venture Partners Menlo Ventures Norwest Venture Partners Tenaya Capital

8/2010

$17

Total Capital Raised:

$32M

Page 109

March 14, 2012

Industry Report

Name: BiTMICRO Networks Headquarters: Fremont, CA CEO: Rey Bruce Phone: 510-623-2341 Year Founded: 1995 Website: bitmicro.com Description: BiTMICRO Networks develops high performance, flash-based solid state disks (SSDs) and storage management technologies. BiTMICRO brings a new dimension to the networked world by delivering a product that becomes the heartbeat of the computing environment. BiTMICRO's flagship product is the E-Disk solid state flash drive, a component-level storage and data retrieval solution. Investors: Woodside Fund Financing History: (US$ in Millions) Total Capital Raised: Date 9/2007 Amount Raised $9.3 $9.3M Investors Woodside Fund

Page 110

March 14, 2012

Industry Report

Name: Cache IQ, Inc. Headquarters: Austin, TX CEO: Joel Trammell Phone: 510-623-2341 Year Founded: 2010 Website: cacheiq.com Description: Cache IQ is a storage solutions company delivering the industrys first and only intelligent caching and visibility product designed to tame the out-of-control growth of network attached storage while optimizing critical application performance. Investors: various Financing History: (US$ in Millions) Total Capital Raised: Date 1/2011 Amount Raised $6 $6M Investors various

Page 111

March 14, 2012

Industry Report

Name: Coraid, Inc. Headquarters: Redwood City, CA CEO: Kevin Brown Phone: 706-548-7200 Year Founded: 2000 Website: coraid.com Description: Coraid Inc., a technology manufacturing company, designs and manufactures computer data storage products. It offers Ethernet SAN storage arrays; SAN management products, such as EtherDrive ESM1000 SAN Manager; WAN gateway products, including EtherDrive WAN Gateway that provides an IP tunnel enabling remote site data backup and disaster recovery; EtherFlash, a solution that provides access to flash drives without the bottlenecks and limitations of legacy storage and Fibre Channel networks; and Caringo-Coraid, a unified storage infrastructure for scaleout block and object storage. The company also provides storage mirroring products, which include EMX EtherDrive mirror appliances; server platforms; and host bus adapters and accessories, such as disk drives and spare parts kits. Its products are used in various applications, such as virtualization, cloud storage, high performance computing, high performance virtual computing, and video/network surveillance applications. The company serves corporations, governmental agencies, educational institutions, and hosting service providers. It sells its products through technology channel partners in the United States and internationally. Coraid Inc. was founded in 2000 and is based in Redwood City, California. The company has product development and manufacturing offices in Georgia. Investors: Allegis Capital, Azure Capital Partners, Inc., Crosslink Capital Ventures, Kinetic Ventures LLC Menlo Ventures, Seagate Technology Financing History: (US$ in Millions) Date 1/2010 Amount Raised $10 Investors Allegis Capital Azure Capital Partners, Inc. Allegis Capital Azure Capital Partners, Inc. Menlo Ventures Allegis Capital Azure Capital Partners, Inc. Crosslink Capital Ventures Kinetic Ventures LLC Menlo Ventures Seagate Technology

11/2010

$25

11/2011

$50

Total Capital Raised:

$85M

Page 112

March 14, 2012

Industry Report

Name: DataDirect Networks Headquarters: Chatsworth, CA CEO: Alex Bouzari Phone: 818-700-7600 Year Founded: 1998 Website: ddn.com Description: DataDirect Networks (DDN) is a provider of data storage and processing solutions and services that enable content-rich and high growth IT environments to achieve the highest levels of systems scalability, efficiency and simplicity. DDN enables enterprises to extract value and deliver results from their information. DDNs customers include the world's leading online content and social networking providers, high performance cloud and grid computing, life sciences, media production organizations and security & intelligence organizations. Deployed in thousands of mission critical environments worldwide, DDN's solutions have been designed, engineered and proven in the world's most scalable data centers to ensure competitive business advantage for today's information powered enterprise. Investors: ClearLight Partners, Iris Capital, Montgomery & Co. Financing History: (US$ in Millions) Date 10/2001 Amount Raised $9.9 Investors ClearLight Partners Iris Capital Montgomery & Co.

Total Capital Raised:

$9.9M

Page 113

March 14, 2012

Industry Report

Name: DDRdrive Headquarters: Paso Robles, CA Phone: 650-804-8227 Year Founded: 2007 Website: ddrdrive.com Description: DDRdrive LLC designs, manufactures, and supplies solid-state storage systems for IOPS intensive applications. The company provides DDRdrive X1, a PCI express expansion card that integrates DRAM and NAND to transform direct-attached enterprise storage. It also offers consulting services. Investors: n/a Financing History: (US$ in Millions) Total Capital Raised: Date Amount Raised n/a n/a Investors

Page 114

March 14, 2012

Industry Report

Name: FlashSoft Headquarters: Sunnyvale, CA CEO: Ted Sanford Phone: 650-262-6663 Year Founded: 2009 Website: flashsoft.com Description: FlashSoft provides software that enables any standard SSD as a high-performance, read-write memory cache in the server, to economically increase application performance and scalability. FlashSoft provides software solutions for Flash Virtualization that eliminate I/O latency to increase application performance, increase VM capability and relieve overhead on storage systems. Its Active Data Management technology uses sophisticated algorithms to observe I/O and cache hot data on server-tier SSD. FlashSoft delivers its solutions directly to customers and in partnership with leading providers of flash memory technology, computing infrastructure and IT solutions. On February 15, SanDisk announced that it had acquired FlashSoft. Terms of the acquisition were not available. Investors: Accelerator Venture Partners, Bullpen Capital, Divergent Ventures, Thomvest Ventures Financing History: (US$ in Millions) Date 12/2010 6/2011 Amount Raised n/a $3 Investors

Accelerator Venture Partners Bullpen Capital Divergent Ventures Thomvest Ventures

Total Capital Raised:

$3M

Page 115

March 14, 2012

Industry Report

Name: Foremay Headquarters: Pasadena, CA Phone: 408-228-3468 Year Founded: 2002 Website: foremay.net Description: Foremay, Inc. is dedicated to designing and developing of Solid State Drives (SSD) for industrial computing, mission critical computing, embedded computing, enterprise servers, web and e-mail servers, commercial laptops, desktops and workstations, and mobile computing. Foremay's vision is to bring high reliability, high ruggedness, high performance and low power consumption ("Green Initiatives") to flash hard drives across the entire storage industry. Foremay's solid state disk benefits your business and systems with high reliability (anti-vibration, anti-shock and antidrop), ultra-fast read/write speed, and high durability storage solutions. Its solid state hard drive also significantly extends the battery life of your mobile computing systems and devices, and significantly reduces your annual power bill amount if you are running servers or RAID arrays. Investors: n/a Financing History: (US$ in Millions) Total Capital Raised: Date Amount Raised n/a n/a Investors

Page 116

March 14, 2012

Industry Report

Name: GreenBytes Headquarters: Ashaway, RI CEO: Robert Petrocelli Phone: 401-315-5580 Year Founded: 2007 Website: getgreenbytes.com Description: GreenBytes, Inc. is a provider of high-performance, energy-efficient inline data deduplication storage solutions. Designed for high availability primary storage and utility storage missions, GreenBytes technology features a next-generation Hybrid Storage Architecture (HSA) that combines SSD performance with magnetic density for the industrys highest levels of performance, scalability and reliability in an easy-to-use, cost-effective package purpose-built to bring the efficiencies of data deduplication beyond data protection and into the primary storage market. Investors: Battery Ventures Financing History: (US$ in Millions) Date 9/2009 9/2010 4/2011 Total Capital Raised: Amount Raised $8 $3.5 $1 $12.5M Investors Battery Ventures n/a n/a

Page 117

March 14, 2012

Industry Report

Name: Greenliant Systems Headquarters: Santa Clara, CA CEO: Bing Yeh Phone: 408-200-8000 Year Founded: 2010 Website: greenliant.com Description: Greenliant is a major supplier of flash memory, solid state storage and controller products to leading industrial, networking, automotive, medical and consumer electronics companies. Investors: spun out from Silicon Storage Technology

Page 118

March 14, 2012

Industry Report

Name: GridIron Systems, Inc. Headquarters: Sunnyvale, CA CEO: Mark Lazar Phone: 408-470-4500 Year Founded: 2007 Website: gridironsystems.com Description: GridIron Systems enables to turn customer's performance-challenged IT environment into the fastest data serving infrastructure. The GridIron TurboCharger appliance, a combination of custom silicon and novel real-time analytics, boosts application performance and scalability by reducing the data access latency by up to 1,000 times without changing applications, database software, servers, storage, or storage networks. Investors: Foundation Capital, Mohr Davidson Ventures, Trinity Ventures LP Financing History: (US$ in Millions) Date 10/2009 Amount Raised $20 Investors Foundation Capital Mohr Davidson Ventures Trinity Ventures LP

Total Capital Raised:

$20M

Page 119

March 14, 2012

Industry Report

Name: InnoDisk Corporation Headquarters: Taipei, Taiwan Phone: 886-2-2696-3000 Year Founded: 2005 Website: innodisk.com Description: InnoDisk is a manufacturer and solution provider of flash storage devices for industrial applications and embedded systems. With the extensive experience in NAND flash memory, InnoDisk is capable of developing series of product with high quality, strong performance, and high reliability as well as cost effective. Investors: n/a Financing History: (US$ in Millions) Total Capital Raised: Date Amount Raised n/a n/a Investors

Page 120

March 14, 2012

Industry Report

Name: iXsystems, Inc. Headquarters: San Jose, CA CEO: Michael Lauth Phone: 408-943-1400 Year Founded: 1991 Website: ixsystems.com Description: iXsystems, Inc. provides open source hardware and software solutions. It offers hardware services, such as racking and deployment, professional server installation, custom server configuration, and high performance computing utilizing open source hardware and software, as well as server leasing and financing. The company also provides software services, including open source development and consultation, operating system installation and configuration, customized software installation, and server configuration and deployment. In addition, it offers custom solutions, such as load balancing and OEM private label solutions. Investors: n/a Financing History: (US$ in Millions) Total Capital Raised: Date Amount Raised n/a n/a Investors

Page 121

March 14, 2012

Industry Report

Name: Kaminario Technologies Ltd. Headquarters: Yokneam, Israel CEO: Dani Golan Phone: 972-72-222-4488 Year Founded: 2008 Website: kaminario.com Description: Kaminario, Inc. provides high performance solid state storage devices. It offers K2 DRAM storage appliance, a performance data appliance to support IOPS and various applications without compromising on high availability; Kaminario storage performance technology; Kaminario K2 Storage Appliance Support; ROI and TCO; and Kaminario K2 storage technology comparison. The companys solutions are applied in databases, such as Oracle and MS SQL Server; data mining and warehousing; business intelligence; operational business intelligence; customer relationship management; and Web. It serves financial, telecommunications, Web service providers, government, insurance, healthcare, and education and research industries. Investors: Globespan Capital Partners, Pitango Venture Capital, Sequoia Capital Israel Financing History: (US$ in Millions) Date 4/2008 5/2011 Amount Raised n/a $15 Investors Sequoia Capital Israel Globespan Capital Partners Pitango Venture Capital Sequoia Capital Israel

Total Capital Raised:

$15M

Page 122

March 14, 2012

Industry Report

Name: Kove Headquarters: Chicago, IL CEO: John Overton Phone: 312-850-3308 Year Founded: 2004 Website: kove.com Description: Kove is a pioneering leader in high performance storage. Kove provides patented, core technology components to solve the most challenging storage and data management needs. Investors: n/a Financing History: (US$ in Millions) Total Capital Raised: Date Amount Raised n/a n/a Investors

Page 123

March 14, 2012

Industry Report

Name: Memoright Corp. Headquarters: Taipei, Taiwan CEO: Ying Tsung Kao Phone: 866-2-2218-3789 Year Founded: 2006 Website: memoright.com Description: Memoright is dedicated to collaborating with customers and partners to develop and manufacture highquality Solid State Drives (SSDs). With proprietary core technologies, Memoright is capable of providing customized solutions and efficient support for various applications. Investors: n/a Financing History: (US$ in Millions) Total Capital Raised: Date Amount Raised n/a n/a Investors

Page 124

March 14, 2012

Industry Report

Name: Nexenta Systems, Inc. Headquarters: Mountain View, CA CEO: Evan Powell Phone: 650-215-2720 Year Founded: 2004 Website: nexenta.com Description: Nexenta Systems, Inc. is an open source enterprise class storage software solutions provider. Nexenta has operations around the world in the United States, Germany, Russia, and India. Further, Nexenta's products are being embedded into a growing set of products and solutions created and sold by hardware vendors, software developers, systems integrators, and resellers. Nexenta products are sold largely via resellers, systems integrators, and hardware Original Equipment Manufacturers (OEMs). Investors: Finaves SA, Javelin Venture Partners, KT Corporation, TransLink Capital LLC, Wti Ventures Financing History: (US$ in Millions) Date 10/2010 Amount Raised n/a Investors Finaves SA Javelin Venture Partners KT Corporation TransLink Capital LLC Wti Ventures

Total Capital Raised:

n/a

Page 125

March 14, 2012

Industry Report

Name: NexGen Storage, Inc. Headquarters: Louisville, CA CEO: John Spiers Phone: 720-287-7550 Year Founded: 2010 Website: nexgenstorage.com Description: NexGen Storage, Inc. provides shared storage systems. Its enterprise class storage systems are used for server virtualization, virtual desktop, business critical application, and analytics solutions. The company sells its products through reseller partners. Investors: Access Venture Partners, Grotech Ventures, Next World Capital Financing History: (US$ in Millions) Date 12/2010 Amount Raised $2 Investors Access Venture Partners Grotech Ventures Access Venture Partners Grotech Ventures Next World Capital

11/2011

$10

Total Capital Raised:

$12M

Page 126

March 14, 2012

Industry Report

Name: Nimble Storage, Inc. Headquarters: San Jose, CA CEO: Suresh Vasudevan Phone: 408-432-6212 Year Founded: 2007 Website: nimblestorage.com Description: Nimble Storage, Inc. provides and delivers storage and application solutions and services. Nimble developed the first converged storage and backup solution that significantly reduces the cost and complexity of storing, accessing, and protecting data for enterprises. Investors: Accel Partners LLC, Artis Capital Management, Lightspeed Venture Partners, Sequoia Capital Financing History: (US$ in Millions) Date 12/2008 Amount Raised $8.3 Investors Accel Partners LLC Lightspeed Venture Partners Sequoia Capital Accel Partners LLC Lightspeed Venture Partners Sequoia Capital Accel Partners LLC Artis Capital Management Lightspeed Venture Partners Sequoia Capital

12/2010

$16

7/2011

$25

Total Capital Raised:

$49.3M

Page 127

March 14, 2012

Industry Report

Name: Nimbus Data Systems, Inc. Headquarters: South San Francisco, CA CEO: Thomas Isakovich Phone: 877-664-6287 Year Founded: 2003 Website: nimbusdata.com Description: Nimbus Data Systems, Inc. develops Internet protocol storage systems, software, and controllers for businesses and data-intensive applications. Its products include MySAN iSCSI Servers, iSCSI target software for Windows; Internet protocol Storage Controllers; and HALO Software, software for Internet protocol storage networking. The companys products are used in data consolidation, Microsoft servers, backup and restore, and specialized applications. Investors: Spring Creek Partners Financing History: (US$ in Millions) Total Capital Raised: Date 11/2003 Amount Raised n/a n/a Investors Spring Creek Partners

Page 128

March 14, 2012

Industry Report

Name: Nirvanix, Inc. Headquarters: San Diego, CA CEO: Scott Genereux Phone: 619-794-5650 Year Founded: 1998 Website: nirvanix.com Description: Nirvanix, Inc. provides Internet storage solutions for media and entertainment markets. It offers Global Storage delivery Network (SDN) that provides policy based Cloud Storage for businesses that demand data security, availability, and enterprise level support; FTP Proxy, an FTP Server that allows users to connect to the SDN through FTP Client enabling uploading, downloading, and other file operations through the FTP Protocol; and hNode, a cloud storage solution. The company also provides Migration Tool, which allows users to automatically migrate their S3 files; and Web Client, which enables users to access and explore data stored on the SDN. In addition, it offers ingestion services. The company was founded in 1998 as Streamload, Inc. and changed its name to Nirvanix, Inc. in July 2007. Investors: Windward Ventures, Valhalla Partners (Wednesday Management), Mission Ventures, Intel Capital, European Founders Fund Management Financing History: (US$ in Millions) Date 9/2007 Amount Raised $12 Investors Mission Ventures Wednesday Management Windward Ventures Intel Capital European Founders Fund Management Intel Capital Mission Ventures Wednesday Management Windward Ventures Intel Capital Mission Ventures Wednesday Management Windward Ventures

12/2007 4/2008 4/2009

n/a n/a $5

10/2010

$10

Total Capital Raised:

$27M

Page 129

March 14, 2012

Industry Report

Name: Nutanix, Inc. Headquarters: Santa Clara, CA CEO: Dheeraj Pandey Phone: 408-520-0520 Year Founded: 2009 Website: nutanix.com Description: Nutanix, Inc. designs and develops converged compute and storage infrastructure for virtualized data centers. Its solutions include data center virtualization, disaster recovery, desktop virtualization, and testing and development. Nutanix enables you to virtualize your data center without requiring a storage area network. The converged compute and storage architecture is a hardware and software solution that delivers a purpose-built building block for virtualization. Investors: Accelerator Venture Partners LLC, Blumberg Capital LLC, Lightspeed Venture Partners, Kholsa Ventures LLC Financing History: (US$ in Millions) Date 1/2010 Amount Raised n/a Investors Accelerator Venture Partners LLC Blumberg Capital LLC Lightspeed Venture Partners Blumberg Capital LLC Lightspeed Venture Partners Blumberg Capital LLC Lightspeed Venture Partner Kholsa Ventures LLC

4/2011

$13.2

10/2011

$25

Total Capital Raised:

$38.2M

Page 130

March 14, 2012

Industry Report

Name: NVELO, Inc. Headquarters: Santa Clara, CA CEO: Jiurong Cheng Phone: 408-365-4170 Year Founded: 2010 Website: nvelo.com Description: NVELO, Inc. develops and sells computer storage software solutions. The company offers Dataplex, a software product that enhances the performance of computers by using solid-state drives (SSD) for existing hard disk drives. It markets solutions through computer OEM's and SSD vendors. Investors: SKS Capital LLC Financing History: (US$ in Millions) Total Capital Raised: Date 7/2011 Amount Raised $6.6 $6.6M Investors SKS Capital LLC

Page 131

March 14, 2012

Industry Report

Name: Pivot3, Inc. Headquarters: Austin, TX CEO: Richard Bravman Phone: 512-807-2666 Year Founded: 2003 Website: pivot3.com Description: Pivot3, Inc. provides converged infrastructure for applications in data protection, digital surveillance, field support, and media markets in the public sector, transportation, education, financial, education, small-to-medium business, and retail vertical segments. It offers failover-protected virtual machines, including vBank, a scale-out SSD/disk storage appliance and multiple VMs that deliver compute and storage resources, as well as support various virtual servers and solid state disk drives to extend storage performance across general business applications and bundles; Storage And Compute Stack for capacity-intensive environments that provide Direct Drive Access for virtual machines, as well as federate hardware resources across Pivot3 appliances; and CloudBank, a scale-out disk storage appliance and single VM; and DataBank, a scale-out disk storage appliance. Investors: InterWest Partners, Lightspeed Venture Partners, Mesirow Financial Private Equity, Inc., Silver Creek Ventures Corp., Focus Ventures, Northleaf Capital Partners Financing History: (US$ in Millions) Date 1/2004 Amount Raised $6 Investors InterWest Partners Lightspeed Venture Partners InterWest Partners Lightspeed Venture Partners InterWest Partners Lightspeed Venture Partners Mesirow Financial Private Equity, Inc. Silver Creek Ventures Corp. Focus Ventures InterWest Partners Lightspeed Venture Partners Mesirow Financial Private Equity, Inc. Silver Creek Ventures Corp. Northleaf Capital Partners

2/2005

n/a

1/2008

$24

3/2010

$25

4/2010 Total Capital Raised:

$4 $59M

Page 132

March 14, 2012

Industry Report

Name: Pure Storage, Inc. Headquarters: Mountain View, CA CEO: Scott Dietzen Phone: 650-290-6088 Year Founded: 2009 Website: purestorage.com Description: Pure Storage, Inc. provides enterprise storage solutions. Its products enhance random I/O-intensive applications, such as server virtualization, desktop virtualization (VDI), database (OLTP, rich analytics/OLAP, SQL, NoSQL), and cloud computing. Investors: Sutter Hill Ventures, Greylock Ventures, Redpoint Ventures, Samsung Venture Investment Corp. Financing History: (US$ in Millions) Date 11/2009 7/2010 Amount Raised $4 $18.4 Investors Sutter Hill Ventures Greylock Ventures Sutter Hill Ventures Greylock Ventures Sutter Hill Ventures Redpoint Ventures Samsung Venture Investment Corp.

8/2011

$30

Total Capital Raised:

$52.4M

Page 133

March 14, 2012

Industry Report

Name: RunCore Innovation Technology Co., Ltd. Headquarters: Changsha Hunan, China CEO: Michael Weng Phone: 86-731-8583-7066 Year Founded: 2007 Website: runcore.com Description: RunCore is engaged in industrial solid array series, product sales and solution service and SSD for consumer separately. It aims to stay ahead in solid storage area of mainland and keep in top five among global solid storage enterprises (it is the fifth one among 127 SSD suppliers at the first season 2010 on Storagesearch.com, a global SSD and solid storage research institute). Investors: Hunan Xiangtou Hi-Tech Venture Capital

Financing History: (US$ in Millions) Total Capital Raised:

Date 7/2009

Amount Raised n/a n/a

Investors Hunan Xiangtou Hi-Tech VC

Page 134

March 14, 2012

Industry Report

Name: Scality, Inc. Headquarters: San Francisco, CA CEO: Jerome Lecat Phone: 650-356-8500 Year Founded: 2009 Website: scality.com Description: Scality has developed a massively scalable storage platform providing significantly easier application deployment at a fraction of the cost of traditional storage. Delivered as a software solution, Scality can be deployed to enable email solutions for cable and telecom operators or as a system to enable and grow cloud storage services at hosting providers or other storage-intensive services. Scality sells direct to Internet service providers, cable operators, fixed-line, mobile operators and hosting companies and through value-added distribution channels. Investors: Credit Agricole Private Equity Capital, Galileo Partners SA, IDInvest Partners SA Financing History: (US$ in Millions) Date 6/2010 Amount Raised $5 Investors Credit Agricole Private Equity Capital Galileo Partners SA Credit Agricole Private Equity Capital Galileo Partners SA IDInvest Partners SA

3/2011

$7

Total Capital Raised:

$12M

Page 135

March 14, 2012

Industry Report

Name: SolidFire, Inc. Headquarters: Boulder, CO CEO: Dave Wright Phone: 720-523-3278 Year Founded: 2010 Website: solidfire.com Description: SolidFire provides cloud computing solutions. The company offers software solution that runs on physical servers to deliver highly scalable block storage. The company also focuses on solid-state drives, which provide performance advantages, and low latency response times across highly varied workloads. Investors: Novak Biddle Venture Partners, Valhalla Partners, New Enterprise Associates Financing History: (US$ in Millions) Date 7/2010 Amount Raised $1 Investors Novak Biddle Venture Partners Valhalla Partners New Enterprise Associates Novak Biddle Venture Partners Valhalla Partners New Enterprise Associates Novak Biddle Venture Partners Valhalla Partners

2/2011

$11

10/2011

$25

Total Capital Raised:

$37M

Page 136

March 14, 2012

Industry Report

Name: StorSimple, Inc. Headquarters: Santa Clara, CA CEO: Ursheet Parikh Phone: 408-550-2300 Year Founded: 2009 Website: storsimple.com Description: StorSimple provides data storage solutions for customers to securely and transparently integrate the cloud into on-premises applications and offers a single appliance that delivers high-performance tiered storage, live archiving, cloud based data protection and disaster recovery, reducing cost by up to 80 percent compared to traditional enterprise storage. Investors: Index Ventures, Redpoint Ventures, Ignition Partners, Mayfield Fund Financing History: (US$ in Millions) Date 6/2009 Amount Raised $8 Investors Index Ventures Redpoint Ventures Ignition Partners Mayfield Fund Index Ventures Redpoint Ventures 6/2011 $10.5 Mayfield Fund Index Ventures Redpoint Ventures Total Capital Raised: $31.5M Ignition Partners

9/2010

$13

Page 137

March 14, 2012

Industry Report

Name: Tegile Systems, Inc. Headquarters: Newark, CA CEO: Rohit Kshetrapal Phone: 510-791-7900 Year Founded: 2010 Website: tegile.com Description: Tegile Systems is developing a new generation of affordable feature-rich storage arrays that are up to 5X faster and require 75% less capacity than standard arrays. IT departments use these Zebi arrays to reduce the cost and increase the performance of demanding virtualization, file share and database applications. Tegiles patent-pending technology underpins these achievements. Tegiles Zebi combines a Copy on Write (COW) file system along with proprietary Metadata Accelerated Storage System (MASS) technology resulting in high performance, high capacity and high reliability at low costs. Investors: August Capital Financing History: (US$ in Millions) Total Capital Raised: Date Amount Raised n/a n/a Investors August Capital

Page 138

March 14, 2012

Industry Report

Name: Texas Memory Systems, Inc. Headquarters: Houston, TX CEO: Holly Frost Phone: 713-266-3200 Year Founded: 1978 Website: superssd.com Description: Texas Memory Systems, Inc. manufactures networking equipment used to build mass storage systems. Texas Memory Systems, Inc. designs and manufactures enterprise data storage and digital signal processing systems. It offers rackmount flash storage, rackmount RAM storage, peripheral component interconnect bus storage, digital signal processing, interfaces, and legacy products. The company serves financial, department of defense and intelligence community, high performance computing, telecom, ecommerce, and healthcare industries. It offers its products through resellers. It provides high-bandwidth, low-latency, I/O-intensive data storage systems, as well as digital signal processor boards. The company markets the RamSan line of solid state disks to commercial and government customers, including News Corporation, Cingular Wireless, Sungard Trading and Risk Systems and Lockheed Martin. Their systems are also sold through resellers and OEMs, including Dell, CDW and Sun Microsystems UK. It partners with companies such as Oracle, IBM, QLogic and Brocade. Investors: n/a Financing History: (US$ in Millions) Total Capital Raised: Date Amount Raised n/a n/a Investors

Page 139

March 14, 2012

Industry Report

Name: Tintri, Inc. Headquarters: Mountain View, CA CEO: Kieran Harty Phone: 650-209-3900 Year Founded: 2006 Website: tintri.com Description: Tintri, Inc. develops purpose-built storage systems for virtual machines. The company offers Tintri VMstore, a virtual machine aware storage appliance that enables enterprises to manage and store virtual machines. It also provides purpose-built solid state drives; and Serial ATA hybrid storage arrays. The company serves various companies in electronics, mobile games, geo-informatics, technology, and higher education industries. Investors: Lightspeed Venture Partners, New Enterprise Associates Financing History: (US$ in Millions) Date n/a Amount Raised n/a Investors Lightspeed Venture Partners New Enterprise Associates Lightspeed Venture Partners New Enterprise Associates Lightspeed Venture Partners New Enterprise Associates

5/2011

n/a

6/2011

$18

Total Capital Raised:

$18M

Page 140

March 14, 2012

Industry Report

Name: Unigen Corp. Headquarters: Fremont, CA CEO: Paul Heng Phone: 510-668-2088 Year Founded: 1991 Website: unigen.com Description: Unigen Corporation engages in the design and manufacture of OEM memory, DC-DC power converter, wired and wireless communication, and flash solutions. It offers memory products, including networking and data storage systems and digital media, as well as DRAM, SRAM, and flash modules; subsystems, including voltage regulator modules, mixed memory and critical components modules, and application specific modules; communication products, including Bluetooth modules, dual-band modules, and digital media and wireless modules; casework, including metal stamping and fabrication; and semi-conductors, including discreet legacy and memory integrated circuits, and multi-chip mixed memory. Investors: n/a Financing History: (US$ in Millions) Total Capital Raised: Date Amount Raised n/a n/a Investors

Page 141

March 14, 2012

Industry Report

Name: VeloBit, Inc. Headquarters: Boxoborough, MA CEO: Duncan McCallum Phone: 978-263-4800 Year Founded: 2010 Website: velobit.com Description: VeloBit is developing software that delivers an order-of-magnitude improvement in storage priceperformance. It is pioneering a new approach that can be used with any Solid State Disk (SSD) to deliver industry-leading performance at a lower cost. It is a performance booster for the storage you already own. With the company, you can deploy SSD without changing your applications or primary storage systems-preserving your investments in data protection and data management. It is easy to install, transparent to applications, and works with any block-based storage systems. Investors: Fairhaven Capital Partners LLC, Longworth Venture Partners Financing History: (US$ in Millions) Date 6/2011 Amount Raised n/a Investors Fairhaven Capital Partners LLC Longworth Venture Partners

Total Capital Raised:

n/a

Page 142

March 14, 2012

Industry Report

Name: Violin Memory, Inc. Headquarters: Mountain View, CA CEO: Don Basile Phone: 650-396-1500 Year Founded: 2005 Website: violin-memory.com Description: Violin Memory, Inc. provides scalable memory appliances for large dataset applications. It offers Violin 1010, a memory appliance that supports scalable memory and high performance storage; DRAM, an ideal platform for high performance caching, indexes, and metadata; vCACHE, a multi-terabyte capacity network file system cache for enterprise applications; and Flash, which improves the economic and performance metrics of flash storage systems for the data center. The companys products are used in the areas of databases and analytics, metadata servers, scientific computing, and storage caching. Investors: Toshiba Corporation, Juniper Networks (Venture Capital) Financing History: (US$ in Millions) Date 4/2010 2/2011 Amount Raised $17 $35 Investors Toshiba Corporation Juniper Networks (Venture Capital) Toshiba Corporation Juniper Networks (Venture Capital) Toshiba Corporation

6/2011

$40

Total Capital Raised:

$92M

Page 143

March 14, 2012

Industry Report

Name: Virident Systems, Inc. Headquarters: Milpitas, CA CEO: Kumar Ganapathy Phone: 408-503-0100 Year Founded: 2006 Website: virident.com Description: Virident Systems, Inc. provides solid state storage solutions for scale-out datacenters and enterprises. It offers solid state disks (SSDs) for data-intensive workloads, such as databases, business analytics, simulation, visualization, and computing applications. The company also provides GreenCloud Storage Software, which provides a framework for a systematic build-out of SSDs in a data center; and Virident FlashMAX MLC, a PCIe storage class memory (SCM) solution. Investors: Artiman Ventures, Spansion, Inc., Globespan Capital Partners, Sequoia Capital, Cisco Systems, Intel Capital Financing History: (US$ in Millions) Date 7/2007 Amount Raised $13 Investors Artiman Ventures Spansion, Inc. undisclosed investors Artiman Ventures Globespan Capital Partners Sequoia Capital Artiman Ventures Globespan Capital Partners Sequoia Capital Cisco Systems Intel Capital

2/2010 11/2010

$3.4 n/a

11/2011

$21

Total Capital Raised:

$38M

Page 144

March 14, 2012

Industry Report

Name: Virtium Technology, Inc. Headquarters: Rancho Santa Margarita, CA CEO: Phu Hoang Phone: 949-888-2444 Year Founded: 1997 Website: virtium.com Description: Virtium Technology, Inc. provides memory and solid state storage solutions for single board computers and industrial systems. The company offers memory modules and storage products, including SSDDR SODIMM, a dual function memory module; LeanSTOR, an AMC storage device designed with flash components soldered directly on board; and memory solutions, including flyers and datasheets. Investors: n/a Financing History: (US$ in Millions) Total Capital Raised: Date Amount Raised n/a n/a Investors

Page 145

March 14, 2012

Industry Report

Name: WhipTail Technologies, Inc. Headquarters: Whippany, NJ CEO: Daniel Crain Phone: 888-550-8136 Year Founded: 2007 Website: whiptailtech.com Description: WhipTail Technologies, Inc. is an independent manufacturer of solid-state storage arrays designed to drastically improve the scalability and performance of applications such as virtualization, databases, financial trading and email. WhipTail is currently a leader in extremely high performance, pure solid state storage arrays for the commercial, enterprise and government markets. WhipTail's storage arrays are installed worldwide, in numerous mission critical virtualization and database environments. Its datacenter and virtual desktop systems dramatically reduce delays related to disk contention and access times thereby allowing database or virtualization servers to process more data in dramatically less time, while lowering the overall TCO by up to 90% when compared to traditional storage. Investors: RRE Ventures LLC, Spring Mountain Capital LP Financing History: (US$ in Millions) Date 7/2011 Amount Raised n/a Investors RRE Ventures LLC Spring Mountain Capital

Total Capital Raised:

n/a

Page 146

March 14, 2012

Industry Report

Name: XtremIO, Inc. Headquarters: San Jose, CA CEO: Ehud Rokach Phone: 408-441-0336 Year Founded: 2009 Website: xtremio.com Description: XtremIO is developing the next generation of Solid-State based storage systems, delivering value to customers through dramatic gains in performance, economics, and maintainability while enabling application acceleration, higher levels of consolidation, operational simplification, and cost reduction. Investors: Lightspeed Management Co. LLC, Battery Ventures, Giza Venture Capital, Jerusalem Venture Partners Financing History: (US$ in Millions) Date 11/2009 Amount Raised na Investors Giza Venture Capital Jerusalem Venture Partners Battery Ventures Giza Venture Capital Jerusalem Venture Partners Lightspeed Management Co. LLC

6/2011

$14

12/2011 Total Capital Raised:

na $14M

Page 147

March 14, 2012

Industry Report

Name: Zetta, Inc. Headquarters: Sunnyvale, CA CEO: Ali Jenab Phone: 650-590-0950 Year Founded: 2007 Website: zetta.net Description: Zetta, Inc. provides enterprise cloud storage services for data protection. It offers enterprise cloud data protection, on-demand cloud storage, and offsite data protection services. The company also provides Zetta Storage Service, an enterprise network storage service; and ZettaMirror, an agent software that works with the Zetta Storage Service to provide offsite data protection for the enterprise data. Investors: Amidzad Partners, Foundation Capital, Sigma Partners Financing History: (US$ in Millions) Date 12/2007 9/2008 Amount Raised n/a $10.7 Investors Amidzad Partners Foundation Capital Sigma Partners Foundation Capital Sigma Partners Foundation Capital Sigma Partners

9/2010

$11.5

9/2011

$9

Total Capital Raised:

$31.2M

Page 148

March 14, 2012

Industry Report

Appendix II: Venture Capital Profiles

Accel Management Co., Inc. Palo Alto, CA accel.com Accel Management Co, Inc. is a private equity and venture capital firm specializing in investments in seed, start-ups, early venture, late venture, and growth stage companies. The firm prefers to invest in cloud computing and storage infrastructure, consumer and business centric Internet, media including digital media, energy, enterprise software and services, consumer internet, internet, healthcare and biotech, social and mobile, technology, networking systems, retail consumer, security, semiconductors, technology enabled services, clean technology, web search, online advertising, SaaS, financial services, and open source software. It prefers to invest in Europe with a focus on Finland, Germany, Ireland, Switzerland, United Kingdom, China, India, Israel, and the United States. The firm also seeks to invest in Brazil and Latin America. Within China, the firm targets light-emitting diodes, the smart grid, clean transmission, and solar photovoltaic material investments. Accel Management Co, Inc. was founded in 1983 and is based in Palo Alto, California with additional offices in London, United Kingdom; New York, New York; New Delhi, India; and Bengaluru, India. Accelerator Venture Partners San Francisco, CA acceleratorventures.com Accelerator Ventures is a venture capital firm specializing in seed stage and early stage investments. The firm considers investments in technology and consumer-facing businesses. It seeks to invest in companies seeking their first round of seed funding. Access Venture Partners Denver, CO accessventurepartners.com Access Venture Partners is a venture capital firm specializing in seed and early stage investments. The firm invests in the technology sector with a focus on clean technology, SaaS, SaaS infrastructure, data integrity, new media, and web commerce. It seeks to invest in businesses in large markets and does not consider narrow niches. The firm prefers to invest throughout North America. It invests between $0.25 million and $2 million. The firm seeks to take its portfolio companies public in a period of two years to four years. Access Venture Partners was founded in 1999 and is based in Denver, Colorado with an additional office in Austin, Texas. Amidzad Partners Redwood City, CA amidzad.com Amidzad Partners is a venture capital arm of Amidi Group specializing in seed, startup, and early-stage emerging growth companies. The firm typically invests in enterprise software and security; business and consumer product and services; communication and storage; and life sciences. It seeks to invest in the United States with a focus on the West Coast. Amidzad Partners was founded in 2000. Allegis Capital Palo Alto, CA allegiscapital.com Allegis Capital is a venture capital firm specializing in incubation and investments in seed and early-stage startup companies. It typically invests in developing enabling technology and infrastructure, which serve emerging information technology markets with a focus on enabling hardware devices, enterprise software solutions, broadband and wireless delivery technologies, wired and wireless communications, digital media and commerce enablers, internet delivery platforms and services, security, and information services. The firm prefers to invest in the western region of United States and selected investments in Asia. It seeks to invest between $3 million and $5 million in its first investment to acquire an initial 20% ownership stake in companies and provides capital support for subsequent financings. The firm seeks opportunities in companies in growth markets with 50-100% compound annual growth rates and greater than 50% gross margins. The firm typically acts as the lead investor and will partner with other venture capital firms. It prefers a board seat

Page 149

March 14, 2012

Industry Report

in its portfolio companies. Allegis Capital was founded in 1995 and is headquartered in Palo Alto, California with additional offices in Santa Monica, California and San Francisco, California. Artiman Ventures East Palo Alto, CA artimanventures.com Artiman Ventures is a venture capital and private equity firm specializing in incubation, seed, early stage, and bridge investments. The firm primarily invests in the technology sector. It seeks to make an initial investment of $100,000 seed to a conventional Series A investment between $2 million and $4 million. The firm prefers to limit aggregate investment over the life of a company to $10 million in capital. It seeks to be the first institutional investor in its portfolio companies. Artiman Ventures is based in East Palo Alto, California with an additional office in Bangalore, India. Artis Capital Management, L.P. San Francisco, CA artiscapital.com Artis Capital Management, L.P. is an employee owned hedge fund sponsor. The firm provides its services to pooled investment vehicles. It invests in the public equity markets of the United States. The firm also makes private equity investments. It invests in a value stock companies. The firm employs a fundamental analysis along with top-down approach to select its sectors and bottom-up stock picking approach to make its investments. It also uses multiple strategies to create its portfolio. The firm typically invests in technology sector. Artis Capital Management is based in San Francisco, California. August Capital Menlo Park, CA augustcap.com August Capital is a private equity and venture capital firm specializing in investments in both early and later stage companies. The firm invests in all stages with a particular focus on late stage, management buyouts, and PIPES. It seeks to invest throughout the information technology spectrum including IT infrastructure, data center technologies, systems management, security, storage, and cloud computing systems and software. The firm prefers to be the lead investor and first professional investors in its portfolio companies and also prefers a board seat. August Capital was founded in August 1995. Azure Capital Partners San Francisco, CA azurecap.com Azure Capital Partners is a venture capital firm specializing in early stage investments. The firm also invests in startups with differentiated products and defensible technologies. These startups develop products based upon a technology that allows them to compete effectively with much larger incumbents and that enables a time to market advantage. The firm seeks to invest in information technology sector with a focus on software; media; and related web properties. It seeks to acquire a board seat in its portfolio companies. Azure Capital Partners was founded in 2000 and is based in San Francisco, California with an additional office in Menlo Park, California. Battery Venture Partners Waltham, MA battery.com Battery Ventures is a private equity and venture capital investment firm specializing in seed, start-up, early and late stage ventures, leveraged buy-outs, acquisitions, growth equity, PIPES, spin-outs, roll-ups, take private transactions, buyouts, special situations, expansion deals, secondary transactions, and distressed situations. It also invests private equity in undervalued companies. The firm seeks to invest in companies of all stages. The firm specializes in equity and debt investments in technology driven companies. It prefers to invest in companies operating in the, Internet; Internet applications; ; digital media; software as a service; healthcare; content management and monetization; data collection and analysis technologies; enterprise IT; big data, analytics, and cloud computing infrastructure; interactive multiplayer gaming; next generation financial products and transaction systems; IT infrastructure; Networking for virtualized and cloud environments, online marketplaces, social media, network monitoring and management, mobile data, data centers,

Page 150

March 14, 2012

Industry Report

wired/wireless service providers, media, financial services and technologies; marketing services; technology enabled businesses; specialty retail; media; education; clean technology; communications infrastructure; analytical instrumentation for laboratory applications; advanced technology; semiconductors and components; security; communication services; and infrastructure technologies. For private equity investments, the firm focuses on companies operating in cleantech, software; financial services; technology enabled businesses; IT services and outsourcing; communications components, systems, and services; and industrial technology sectors. It also invests in, cable infrastructure; network and wireless infrastructure; security software and services; computing and storage infrastructure; application software; e-commerce products and services; consumer technology; energy; and materials industries. The firm seeks to invest in companies based in United States; Canada; China; Europe; North America; India (focused on the, consumer finance; consumer internet; mobile; software; technology enabled services; BPO / ITES; components, infrastructure and communications services; and clean technology sectors); and Israel (focused on the, advanced materials and clean technology; communications services and infrastructure; financial services and technology enabled businesses; and semiconductors and components sectors). It primarily invests between $5 million and $50 million over the life of the investment and has the ability to invest up to $100 million. The firm invests up to $50 million in seed and early stage investments or more in late stage or expansion investments. In case of private equity investments the firm can invest over $100 million equity in companies with revenues between $20 million and $200 million with an EBITDA from negative to $25 million. In case of investments in India, the firm invests $1 million to $10 million in early stage investments and between $10 million and $30 million in growth stage investments. It primarily acts as the lead institutional investor in its investments. The firm seeks to take board membership in its portfolio companies. Battery Ventures was founded in 1983. Blumberg Capital San Francisco, CA blumbergcapital.com Blumberg Capital is a venture capital firm specializing in investments in seed and early stage companies. It seeks to invest in emerging information technology companies with a focus on digital & social media, SaaS/technology enabled services, mobile technology and internet companies. The firm prefers to invest in private companies based in United States and Israel. However it is also receptive to international opportunities within its focus sectors. The firm invests between $0.1 million and $2.5 million in seed and Series A rounds and can participate in follow-on rounds up to $6 million per company. It prefers to lead or co-lead early stage investments and seeks to invest at least for 12 to 18 months. Blumberg Capital was founded in 1991 and is based in San Francisco, California with an additional office in Israel. Bullpen Capital Menlo Park, CA bullpencap.com Bullpen Capital is a venture capital firm. It seeks to invest in technology companies seeded by super angel investors. The firm focuses on leading second round investments. Bullpen Capital is based in Menlo Park, California. ClearLight Partners Newport Beach, CA clearlightpartners.com ClearLight Partners is a private equity firm specializing in middle market leveraged buyouts, management-led industry consolidations, recapitalization of family-owned businesses, and growth equity investments. It does not invest in venture or early stage investments. The firm targets investments in sectors including specialty manufacturing and distribution service and consumer products and services. Within services business sector, it invests in business services, education and training, healthcare services, specialty financial services sectors. For business service sector, it makes investments in business process outsourcing, human resource outsourcing, marketing services, third party administrators, third party logistics, and security services. In consumer products and services sector, the firm mainly invests in specialty foods and supplements, wellness and weight loss products and services, multi-level marketing and direct selling, companies serving the Hispanic community, childrens products and specialty or educational toys, and pet products and services. For specialty financial services, it focuses on collections and receivables management, specialty finance, and electronic payments and transaction processing. Within education and training sector, the firm prefers to invest in service providers to post-secondary colleges, assessment, tutoring, and test or certification preparation businesses, supplemental publishers, and corporate training. For healthcare services sectors, it makes investments in personal emergency response systems, dental laboratories, and contract research and sales. In specialty manufacturing and distribution sector, the firm

Page 151

March 14, 2012

Industry Report

targets investments in aftermarket performance automotive, specialty manufacturing and distribution, and industrial consumables. The firm primarily invests in companies based in United States and Canada. It seeks to invest between $10 million and $50 million of equity capital in companies with revenues between $20 million and $250 and EBITDA between $5 million and $25 million. The firm makes investments in the form of cash, notes, stock or contingent payments or any combination and can also provide mezzanine financing with its equity investment. It prefers to lead larger investments with co-investors, making control investments or acquiring a board seat in its portfolio companies. The firm may also consider minority positions in its portfolio companies. It typically exits its investments within a period of five to seven years. ClearLight Partners LLC was founded in 2000 and is based in Newport Beach, California. Credit Agricole Private Equity Paris, France ca-privateequity.com Credit Agricole Private Equity is a private equity arm of Credit Agricole S.A. The firm specializes in direct and fund of fund investments. Within direct investments, it specializes in seed, startup, early venture, development stage, mid venture, late venture, emerging growth, growth capital, replacement capital, middle market, successions, leveraged buyout, mezzanine, PPP infrastructure, co-investments, mezzanine, special situations, recapitalization, industry consolidation, mature, turnaround, and investing in small and medium enterprises. The firm prefers to invest in all the sectors including the information technology, life sciences, and renewable energy sectors. Within information technology, the firm invests in electronics, materials, components, systems, laboratory and manufacturing equipment, design and test software, telecoms, mobile, fiber optics, networks and operators, software applications, development, administration and infrastructure, web-based technology services, media, and e-commerce. In life sciences, the firm invests in drug development, diagnostics, and medical device. Under renewable energy, the firm invests in wind farms, biomass, hydraulic, photovoltaic, thermal solar, geothermal, biogas, and fuel cells. It invests in companies based in Europe and North America region with a focus on France and Italy. The firm makes direct private equity investment in non-listed companies. It seeks to invest between 1 million ($1.32 million) and 50 million ($65.91 million) in companies having maximum revenues up to 250 million ($329.55 million). While co-investing, the firm seeks to take minority stakes in its portfolio companies. It prefers to have a board seat on its portfolio companies. The firm seeks to invest for a period between two years and six years. Crdit Agricole Private Equity was founded in 2000 and is based in Paris, France with additional offices in Lyon, France; Philadelphia, Pennsylvania; and Milan; Italy. Crosslink Capital, Inc. San Francisco, CA crosslinkcapital.com Crosslink Capital, Inc. is a private equity and venture capital firm specializing in seed, early, mid, late venture investments and growth equity in private and public companies and in small to midcap public equities and public growth companies. The firm also provides bridge financings. It primarily invests in energy technologies, cloud computing, financial technologies, Software-as-a-Service, Enterprise IT, communication services and infrastructure, communications, computing and semiconductors, digital media and Internet services, IT hardware, consumer, and software and business services. The firm also invests in technology, information/data, and financial services sectors and through the Alpha Seed Program typical investment sizes range from $100,000-500,000. It manages hedge funds for its clients. The firm invests in the public equity markets. It employs long/short strategy to make its investments in mid cap stocks of companies. The firm prefers to be the lead investor in transactions and takes a board seat in its portfolio companies. Crosslink Capital, Inc. was founded in 1989. Divergent Ventures, LLC Seattle, WA divergentventures.com Divergent Ventures, LLC is a venture capital firm specializing in early stage investments. The firm prefers to invest in software, Internet infrastructure, and information technology enabled services. Divergent Ventures, LLC is headquartered in Seattle, Washington.

Page 152

March 14, 2012

Industry Report

European Founders Fund Management GmbH Munich, Germany europeanfounders.com European Founders Fund Management GmbH is a venture capital firm specializing in seed, early, growth, and later stage ventures. The firm focuses on investing in the equity of young technology companies. It prefers to invest in Internet, software, and wireless projects. The firm primarily invests in companies based in Europe, United States, and China. It typically invests between 500,000-5M. The firm invests between 100,000-1M during the seed stage, 1-3M during the early stage, and 3-8M during the growth stage. It prefers to be the lead investor and have a Board seat in its portfolio companies. European Founders Fund Management GmbH was founded in 2007. Fairhaven Capital Partners, LLC Cambridge, MA fairhavencapital.com Fairhaven Capital Partners, LLC is a venture capital firm specializing in investments in early stage, start-up stage, and growth stage companies. The firm makes investments in the technology sector. It focuses on enabling technologies and platforms that may become the foundation of new Internet and mobile products and services and drive value to enterprises and SMBs; digital media infrastructure that drives value for digital content producers and owners; financial services platforms that will be adopted by institutions and consumers; disruptive materials technology; information technology and physical security products and services. It primarily invests in North America. Fairhaven Capital Partners, LLC was founded in 2007 and is based in Cambridge, Massachusetts. Finaves SA Barcelona, Spain finaves.com Finaves S.C.R., S.A. is a venture capital firm specializing in seed, start-up, early venture, and mezzanine investments. The firm typically invests in new technology and Internet, although it may consider investments in all sectors. It gives priority to business projects drawn up by students and alumni of the IESE Masters Programs. The firm invests in companies based in Europe, North and South America, and Spain, but will consider all geographical investments. It will invest a maximum of 300,000 ($383,300) in each investment. The firm takes a minority stake in the portfolio companies and seeks to take a seat on the Board of its portfolio companies. It seeks to exit its investments between four years and six years. Finaves S.C.R., S.A. was founded in 2000 and is headquartered in Barcelona, Spain. Focus Ventures Palo Alto, CA focusventures.com Focus Ventures is a venture capital firm specializing in expansion stage and emerging growth technology companies. It typically invests in privately held companies that seek additional capital for expansion or to execute strategic acquisitions. The firm seeks to invest in software and communications companies that have completed initial product development and have begun the marketing of their products and/or services. It prefers to invest between $3 million and $8 million. The firm seeks to be the lead investor in 75 percent of the rounds in which it participates. It prefers to invest in technology companies that have previously received backing from a leading early stage venture capital firm. Focus Ventures was founded in 1997 and is based in Palo Alto, California. Foundation Capital Menlo Park, CA foundationcapital.com Foundation Capital is a venture capital firm that invests in early-stage information technology companies. The firm prefers to invest in startups in their formative stages. It will consider investments in the following sectors: enterprise software and on-demand services; networking, storage and telecommunications; fables semiconductor and EDA; cleantech; consumer electronics; data communication; energy management; digital media; internet infrastructure and technologies; financial services; business-to-consumer e-commerce; internet media, and data networks. Within cleantech, the firm invests in energy efficiency and intelligence, green building materials and industrial processes, food quality, clean water, alternative energy and storage, advanced materials, carbon capture, and green consumer services and marketplaces. It prefers to invest in India and United States. The firm typically makes an initial investment between $1 million and $10 million and

Page 153

March 14, 2012

Industry Report

participates in each subsequent round of financing. It prefers to act as the lead investor and takes a board seat in its portfolio companies. Foundation Capital was founded in 1995 and is headquartered in Menlo Park, California. Galileo Partners SA Paris, France galileo.fr Galileo Partners is a private equity and venture capital firm specializing in investments in seed, early, mid, and late venture, expansion, turnarounds, spin-offs, leverage buyouts, and IPO stage. The firm does not invest in biotechnologies. It invests in energy, Information technology, Internet, software, media, electronics, semi-conductors, telecommunications, services, medical, medical technology, and components sectors. It targets companies based in Europe with a focus on French companies. The firm initially invests between 1 million ($1.49 million) and 5 million ($6.37 million) in first round with total investments of 10 million ($14.94 million), funded over several rounds. It seeks to invest as a sole investor and as a lead and co-investor in syndicated venture financings with international and French investors. The firm prefers to have a seat on the companys board of directors and acquires minority stakes in companies. It also takes a majority stake in some companies. Galileo Partners was founded in 1989 and is headquartered at Paris, France. Giza Venture Capital Tel-Aviv, Israel gizavc.com Giza Venture Capital is a venture capital firm specializing in seed through late stage, emerging growth, and growth capital investments. It seeks to invest in technology companies with a focus on communications, semiconductors, information technology, enterprise software, new media, Internet, life sciences, cleantech, and gaming and entertainment. The firm supports Israeli related portfolio companies to expand and partner in Taiwan and the Greater China region and also invests in Polish companies. It seeks to invest between $300,000 million and $8 million in its portfolio companies. The firm seeks a seat on the Board of Directors of its portfolio companies. Giza Venture Capital was founded in 1992 and is headquartered in Tel Aviv, Israel with an additional office in Singapore, Singapore. Globespan Capital Partners Boston, MA globespancapital.com Globespan Capital Partners is a venture capital firm specializing in investments in seed, emerging growth, growth capital, early, middle, and late stage companies. The firm also provides incubation services. It typically makes investments in companies engaged in communications, software, information technology, cleantech, internet infrastructure, and storage and systems peripherals. The firm prefers to invest globally. It seeks to invest between $4 million and $6 million in its portfolio companies. The firm acts as a lead investor and seeks to acquire a board seat in its portfolio companies. Globespan Capital Partners was founded in 2003 and is based in Boston, Massachusetts with additional offices in Palo Alto, California. Grotech Ventures Vienna, VA grotech.com Grotech Ventures is a venture capital firm specializing in investments in start-ups, early stage, and mid venture stage of investments. It typically makes investments at all stages of a companys growth with a bias to invest early. The firm does not invest in life sciences and biotech sectors and in foreign entities. It targets investments in communications and information technology infrastructure, networking, software and services, and Internet and digital media. The firm makes venture capital investments in the information technology sector with a focus on VOIP, IP convergence and Ethernet, wireless, storage, software-as-a-service, information security, Internet commerce, healthcare information technology, technology outsourcing, and data center convergence. It targets investments in underserved venture markets in the United States including the Mid-Atlantic, Southeast, and Colorado. The firm initially invests between $500,000 and $5 million in its portfolio companies. It prefers to be a lead investor and acquires a board seat in its portfolio companies. The firm seeks to make both control and minority equity situations and may also co-invest to execute transactions requiring larger equity commitments. Grotech Ventures was founded in 1984 and is based in Vienna, Virginia with an additional office in Hunt Valley, Maryland.

Page 154

March 14, 2012

Industry Report

GreyLock Partners Menlo Park, CA greylock.com Greylock Partners is a venture capital firm specializing in seed/start up, early venture, mid venture, late venture, growth, follow-on, and late stage investments. The firm invests in all areas of technology including consumer and services; enterprise infrastructure; enterprise software; biotechnology; communications; and semiconductor. Within consumer and services, it seeks to invest in advertising, commerce, gaming, media, social, financial, healthcare, internet, mobile, retail, and systems integration. For its enterprise infrastructure investments, the firm seeks to invest in compute or data center, internet technology, management, networking, security, and storage. For enterprise software, it focuses its investments in applications and software-as-a-service (SaaS). Within semiconductor, the firm invests in EDA, networking, and wireless. It prefers to invest in companies based in North America, United States, India, China, Europe, and Israel. It seeks to invest between $0.05 million and $0.5 million in seed stage companies, $1 million to $10 million in early stage companies. It typically invests $25 million and $200 million in other companies. The firm prefers to take a board seat on its portfolio companies. Greylock Partners was founded in 1965 and is based in San Mateo, California with additional offices in Cambridge, Massachusetts; Bangalore, India; Beijing, China; and Herzliya Pituach, Israel. Hunan Xiangtou Hi-Tech Venture Capital Co., Ltd. Changsha, China hnhvc.com Hunan Xiangtou Hi-Tech Venture Capital Co., Ltd. is a venture capital firm specializing in startups and growth capital. It prefers to invest in information engineering, new materials, optical electromechanical integration, biology, and medicine. Hunan Xiangtou Hi-Tech Venture Capital Co., Ltd. was founded in 2000 and is based in Hunan, China. IDInvest Partners SA Paris, France idinvest-partners.com IDInvest Partners, formerly known as AGF Private Equity, S.A., is an investment firm specializing in direct investments and fund of funds investments. For direct investments, it focuses on venture capital investments and finances the creation and development of young companies and invests primarily in the first or second venture capital financing rounds. On a case by case basis, the firm also invests in either younger companies (seed capital) or more mature companies. It seeks to invest in the information technology (software, Internet, telecommunications, networks, hardware, and components) and life sciences (biopharmacy, diagnostics, and medical instruments) sectors. For fund of funds investments, the firm targets primary funds, secondary transactions, and co-investments in French funds. It participates in leverage buyout coinvestment operations on a case-by-case basis and also makes secondary acquisitions of stakes in funds or asset portfolios. It concentrates its investments in leverage buyout funds. However, the firm diversifies its investments moderately through venture capital funds in order to benefit from research and development into new products in Europe. The firm prefers to invest in companies headquartered in European Union, even if their commercial activities extend to other geographical areas. It typically invests between 1 million ($1.5 million) and 50 million ($74.74 million) in each round, taking a minority interest in its portfolio companies with revenues up to 500 million ($670.44 million). The firm has the capacity to invest up to 10 million ($7.58 million) per company over several rounds of financing. It acts as the leader or co-leader in its investment operations and generally takes a seat on the board of directors or the supervisory board of its portfolio companies. IDInvest Partners was founded in 1995 and is based in Paris, France. As of May 31, 2010 it operates as a subsidiary of Groupe IDI. Ignition Partners Bellevue, WA ignitionpartners.com Ignition Partners is a venture capital firm investing in seed and early-stage companies. The firm typically invests in communications, Internet, software, and services across business and consumer targets. It often takes lead positions in funding rounds in which it participates. The firm prefers to take a board seat in its portfolio companies. Ignition Partners was founded in 2000 and is based in Bellevue, Washington with additional offices at Shanghai, China, Hong Kong, Hong Kong, and Beijing Province, China.

Page 155

March 14, 2012

Industry Report

Index Ventures Geneva, Switzerland indexventures.com Index Ventures is a venture capital firm specializing in seed stage, very-early stage, early stage, mid stage, and growth stage start-ups, series A, and series B rounds. It makes investments in information technology, medical devices, and life sciences companies in green technology, biotechnology, clean technology, consumer and Internet infrastructure, enterprise software, software, mobile software, and therapeutic companies. Under therapeutic areas, the firm focuses on oncology, CNS, immunology, and inflammation. It invests in companies located in Europe with a focus on Switzerland and Israel, but can co-invest with local venture capitalists outside of Europe. The firm also makes investments in the United States. The firm typically invests between $3 million and $10 million and can sometime invest as little as $500,000 in seed rounds. It may also invest $1 million or $2 million in very-early stages. The firm seeks to invest between $5 million and $20 million in a company over the life of an investment. It targets investments in market segments with annual sales in excess of $1 billion. The firm seeks to have a board seat and does not seek control its portfolio companies through ownership. Index Ventures was founded in 1996 and is based in Geneva, Switzerland with additional offices in San Francisco, California; London, United Kingdom; and St. Helier, Channel Islands. InterWest Partners Menlo Park, CA interwest.com InterWest Partners is a venture capital firm specializing in investments in early, mid, and late venture stages and PIPES. It makes ongoing capital investments throughout the full range of venture investment stages including development stages. The firm seeks to invest in the information technology companies with a specific focus on systems, enterprise software and services, wireless and mobility, digital media and Internet, datacenter infrastructure, communications, semiconductors, and components; and life sciences companies with specific focus on biopharmaceuticals, drug discovery and tool discovery, medical devices and diagnostics, service, and healthcare information technology companies. Its investments range from $7 million to $15 million over several rounds of financing. The firm prefers to be the lead investor and typically serves on the Board of Directors of its portfolio companies. It generally holds its investments in portfolio companies for a period of five years and in some cases over a period of ten years or more. InterWest Partners was founded in 1979 and is based at Menlo Park, California with an additional office in Dallas, Texas. Iris Capital Paris, France iriscapital.com Iris Capital is a private equity and venture capital firm specializing in early stage, late stage, growth capital, acquisition, and developmental capital investments. It also invests in middle market and mature companies. The firm prefers to invest in special situations such as public-to-private or secondary transactions. It focuses on media and entertainment, communications, electronics, and information technology sectors. Within media and entertainment, the firm invests in TV production and broadcasting, music, video games, cinema, publishing, advertising, and technical services and solutions. In communications, it seeks to invest in value-added service providers, telecoms, fixed and wireless operators, equipment, and software. Within IT, the firm invests in enterprise software, infrastructure and storage, security, IT services and ASPs, new technologies, and Internet Services sectors. It primarily invests in companies based in Europe with a focus on Continental Europe (France, Germany, Benelux, and Switzerland) and North America. It typically invests between 3 million ($4.23 million) and 8 million ($11.27 million) per round and can invest up to 20 million ($28.17 million) over the life of the company with sales values between 5 million ($6.70 million) and 100 million ($134.08 million). The firm typically acts as a lead or co-lead investor in European investments and prefers to partner with a local fund as a lead investor for North American investments. It seeks to take a board seat in its portfolio companies and takes a minority interest in the companies. Iris Capital was founded in 1986 and is based in Paris, France with an additional office in North Rhine-Westphalia, Germany. Javelin Venture Partners San Francisco, CA javelinvp.com Javelin Venture Partners is a venture capital firm specializing in seed, start-up, and early stage investments. It seeks to invest in the technology sector with a focus on digital and social media; web services and commerce; and mobile, cloud

Page 156

March 14, 2012

Industry Report

and enterprise or healthcare IT. The firm seeks to invest globally. It typically invests between $0.5 million and $3 million in the first investment round. Javelin Venture Partners is based in San Francisco, California. Jerusalem Venture Partners Jerusalem, Israel jvpvc.com Jerusalem Venture Partners is a venture capital firm specializing in early stage and startup investments. The firm prefers to invest in enterprise software and digital media technologies including gaming and virtual worlds, mobile media, software and hardware applications, Internet advertising, semiconductors and innovative materials, and communications and networking sectors. It typically invests in companies based in North America, Europe, Israel, and Asia. Jerusalem Venture Partners was founded in 1993 and is based in Jerusalem, Israel with additional offices in New York, New York; London, United Kingdom; Tokyo, Japan; and Shanghai, China. Kholsa Ventures Menlo Park, CA khoslaventures.com Khosla Ventures is a venture capital firm specializing in seed to late stage venture investments, emerging growth, and growth capital. The firm seeks to invest in sectors such as mobile; life sciences; computing; information technology; personalized medicines; alternative energy; cloud services; internet; silicon technology; high efficiency engines, materials; and clean technology areas such as bio-refineries for energy and bioplastics, solar, battery, and other environmentally friendly technologies. It seeks to invest in China. The firm typically invests between $100,000 and $20 million in its portfolio companies. It seeks to have a board seat in its portfolio companies. Khosla Ventures was founded in 2004 and is based in Menlo Park, California. Kinetic Ventures L.L.C. Chevy Chase, MD kineticventures.com Kinetic Ventures, L.L.C. is a venture capital firm specializing in early stage investments. It also considers investments in the seed, growth, and late stages companies in special situations. The firm typically invests in communications, information technology, and power and clean technology sectors. In communications sector, it focuses on last mile connectivity, video over the Internet, quadruple play, wireless infrastructure, telephony over the internet, RFID and sensor networks, handheld delivery, broadcast TV or video on cell phones, and triple play companies. In information technology sector, the firm seeks to invest in mobile platforms, open source, on-demand computing, customized content and entertainment, lite applications, and security and identity management companies. In power and clean technology sector, it focuses on companies having ongoing and rapid technological evolution in communications, energy, and information technologies. The firm prefers to invest in companies based in the United States and selectively in Canada and outside of North America. In selected cases it considers investments in foreign companies. It seeks to invest between $2 million and $7 million in a company and for larger financing transactions, the firm arranges a syndicate of investors. The firm prefers to serves on a portfolio companys Board of Directors. It typically exits its investments through an IPO or acquisition. The firm was formerly known as Arete Ventures. Kinetic Ventures, L.L.C. was founded in 1983 and is based in Chevy Chase, Maryland with an additional office in Atlanta, Georgia. Lightspeed Venture Partners Menlo Park, CA lightspeedvp.com Lightspeed Venture Partners is a venture capital firm specializing in early stage and expansion stage investments. The firm structures and participates in equity and debt financing for start-up and growth companies and also provides incubation services. In China, the firm seeks to invest in early stage and late stages businesses. It prefers to make investments in mobile, social media, software, commerce, gaming, consumer web services, datacenter infrastructure, enterprise automation, digital media and consumer electronics, communications infrastructure technologies, networking, wireless, Internet media and commerce, Internet and e-commerce, semiconductors, cleantech, and other services. In India, the firm seeks to invest in advertising and media, business services, financial services, healthcare, education, and retail. It primarily invests in the United States, Israel, China, and India. The firm seeks to makes investments not exceeding $20 million. It invests between $10 million and $25 million in growth stages in India. The firm invests between

Page 157

March 14, 2012

Industry Report

$2 million and $10 million in early stages in India. It seeks to take a board seat in its portfolio companies. Lightspeed Venture Partners was founded in 1971 is based in Menlo Park, California with additional offices in Beijing, China; Shanghai, China; New Delhi, India; and Herzliya Pituach, Israel. Longworth Venture Partners Waltham, MA longworth.com Longworth Venture Partners is a venture capital firm specializing in seed, early stage, late stage, growth stage, and Series A or Series B financing in emerging technology companies. It primarily invests in technology markets with a focus on internet and digital media, business internet, mobile, systems and infrastructure. The firm seeks to invest in companies in New England, Mid-Atlantic, West Coast and elsewhere. It seeks to invest between $250,000 to $4 million in its portfolio companies but the firms typical investments ranges between $2 million and $3 million in the first or second round of financing. It seeks to invest in companies which have a potential to grow to more than $100 million in annual revenue. The firm acts as the lead investor in a companys first and second round of funding and can also co-invest. It seeks to take a board seat in its portfolio companies. Longworth Venture Partners was founded in 1999 and is based in Waltham, Massachusetts. Mayfield Fund Menlo Park, CA mayfield.com Mayfield Fund is a venture capital and private equity firm specializing in incubation, seed, series A, early stage, later stage, growth stage, and mid-market investments. The firm does not invest in publicly traded companies and in companies that are not technology-based. It seeks to invest in technology companies with a focus on consumer, enterprise, energy tech, healthcare, communication service providers, and semiconductors chips and components sectors. Within enterprise sector, the firm invests in online marketing and advertising, business application, infrastructure, software, and equipment companies. Within semiconductors chips and components sectors, it invests in communications and application-specific devices. Within energy tech sector, the firm focuses on energy efficiency, management, and cleantech opportunities. It also seeks investments in business services, telecom, mobile, gaming, digital media, engineering services, consumer internet services, interactive entertainment, retail, power ancillary, media, networking components and systems, mobile communications, and business-to-business e-commerce. The firm seeks to invest in companies located across the U.S. with a focus in California as well as cross-border companies domiciled in Europe, India, and China. It also focuses on companies with substantial operations in India with a focus on technology, technology enabled services, consumer services, infrastructure ancillaries, healthcare services, and manufacturing opportunities. The firm also invests in early stage companies with substantial operations in China with a focus on Internet, semiconductor, wireless, new media opportunities, entertainment and advertising services, and e-commerce sectors. It typically makes initial investments between $1 and $3 million. The firm also generally participates in follow-on financing rounds and assists in finding co-investors in these rounds. It prefers to co-invest with other local investors outside California. The firm prefers to lead investment rounds, but also invest in a company even if it is not the lead or co-lead investor. It also seeks to be the first institutional investor in its Chinese investments. Mayfield Fund was founded in 1969 and is based in Menlo Park, California with an additional office in Mumbai, India. Menlo Ventures Menlo Park, CA menlovc.com Menlo Ventures is a venture capital firm specializing in investments in early stage, expansion capital, emerging growth, and seed stage companies. The firm also seeks to invest in pre-IPO stages of a company. It does not invest in companies operating in the life sciences, health care, e-commerce, consumer goods, and retail sectors and in foreign entities. The firm prefers to invest in companies engaged in the connected software and services; social web; software; communications; internet; semiconductor; financial technology; infrastructure; datacenter technologies; application software; mobile software; handset technology; and information technology sector with a focus on areas including internet infrastructure; data storage; consumer related internet sites; and computer hardware sectors. The firm seeks to invest in companies based in the United States. It typically invests from $5 million at the seed stage up to $20 million or more at the late stage and can also invest above $30 million in some opportunities. Menlo Ventures was founded in 1976.

Page 158

March 14, 2012

Industry Report

Mesirow Financial Private Equity, Inc. Chicago, IL mesirowfinancial.com Mesirow Financial Private Equity is an investment arm of Mesirow Financial Inc. specializing in direct equity and fund of funds investments. Through its direct investment, the firm specializes in venture capital, early stage, emerging growth, growth capital, small and middle-market buyouts, and large buyouts. In venture capital, the firm seeks to invest in both early and later-stage growth equity investments in a broad range of healthcare and technology-focused industries. It seeks to invest 30 percent to 40 percent of total committed capital in venture capital. For small and middle-market buyouts, the firm seeks to invest in small to middle-market buyouts, recapitalizations, and industry consolidations of mature companies targeting 40-50% of total invested capital. For large buyouts, it invests in large buyouts, recapitalizations, and industry consolidations of companies with revenues exceeding $500 million targeting 20-25% of total invested capital. Within partnership investments, it seeks to invest in venture capital, buyout, special situations, and international partnerships. In venture capital, the firm seeks to invest in funds focused on early stage information technology and also healthcare investments. It also considers later stage venture capital funds on a select basis. The firm seeks to invest 40 percent of total committed capital in venture capital partnerships. The firm also invests in select larger buyout funds and seeks to invest 35 percent of total committed capital in buyout partnerships. For special situations, it typically invests in mezzanine, distressed/restructuring, and industry focused partnerships with a focus on equity oriented mezzanine sponsors, distressed company investors with control orientation and operational capabilities for turnaround management, and specialist investors with expertise in energy or other asset based industries. The firm targets 15% of total capital commitments for special situations partnerships. The firm prefers to invest 15% of total capital commitments in international partnerships. The firm prefers to invest in companies based in the United States and Western Europe. The firm also seeks to invest in China, India, Israel and Latin America. For international partnerships, it invests in private equity managers in developed economies primarily in Western Europe based middle market private equity managers outside of the United States. It typically invests $3-7 million in direct equity per transaction in companies with annual revenues of $100-150 million and with at least two years of operating experience. The firm seeks to co-invest in its portfolio companies. The firm typically holds its investment for a period of three to four years. Mesirow Financial Private Equity was founded in 1982. Mission Ventures San Diego, CA missionventures.com Mission Ventures is a venture capital firm specializing in early stage investments in emerging markets in software and information technology companies. It does not consider investments in healthcare sector. The firm makes investments in communications, enterprise applications, infrastructure, and technology-driven services. It targets investments in entrepreneurial companies located in Southern California. The firm prefers to invest $2-10M. It seeks to be the lead investor and often takes a Board seat in its portfolio companies. The firm can provide additional funds for follow-on financings and can partner with other venture capital firms as well. Mission Ventures was formed in 1997 and is based in San Diego, California with an additional office in Santa Monica, California. Mohr Davidow Ventures Menlo Park, CA mdv.com Mohr Davidow Ventures is a venture capital firm specializing in seed, startups, and early stage investments. The firm typically invests in information technology, energy and materials, and life sciences companies. Within information technology it invests in Internet services including developing and marketing web-enabled applications and services, Internet media, digital media, consumer Internet, mobiles, semiconductors including semiconductor design, process technology, tools and related hardware, software infrastructure, and software and systems including systems and applications development and systems engineering, and emerging technology. Within energy and materials, it invests in cleantech including energy storage, solar technologies, smart grid, energy management, de-carbonization, and resource supply solutions, renewable, nanotechnology, computational techniques, device technologies, electricity value chain, industrial processes, and manufacturing. Within life sciences it invests in personalized medicine, molecular biology, chemistry, computer science, optics, signal processing, and MEMS companies. It invests in the Silicon Valley, MidAtlantic, and Pacific Northwest regions in the United States. Mohr Davidow Ventures was founded in 1983.

Page 159

March 14, 2012

Industry Report

Montgomery & Co., LLC Santa Monica, CA monty.com Montgomery & Co., LLC is a boutique investment-banking firm that provides financial advisory services to small and middle market public and private companies. The firm provides mergers and acquisition, divestitures, institutional private placement of equity, mezzanine debt or senior debt, negotiation assistance, fairness opinions, due diligence, and strategic advisory services. The company focuses on information technology, communications, media, defense, healthcare, and business and consumer services industries. It has a strategic alliance with Mitsubishi Securities. The company also makes later-stage private equity investments through its investment arm, Digital Coast Ventures. Montgomery & Co., LLC was formerly known as Digital Coast Partners. The company was founded in 1999 and is based in Santa Monica, California with additional offices in San Francisco, California and Kirkland, Washington. New Enterprise Associates Menlo Park, CA nea.com New Enterprise Associates is a venture capital firm specializing in investments in venture growth equity at all points of developments with a 60 percent focus on seed, start-up, and early stage through IPO. The firm opportunistically invests in expansion and mezzanine financing. The firm prefers to invest in the information technology sector with a focus on consumer Internet, consumer technology, consumer mobile, enterprise infrastructure, enterprise services, communications, software and services, digital media, and electronics industries; healthcare sector with a focus on life sciences, biopharmaceuticals, medical devices, healthcare services, and healthcare information technology industries; and clean energy, and energy technology. It invests across the world with a focus on United States, China, and India. New Enterprise Associates was founded in 1978 and is based in Menlo Park, California with additional offices across United States, India, and China. Next World Capital San Francisco, CA nextworldcap.com Next World Capital (NWC) is a venture capital firm specializing in investments in growth stage companies. The firm also invests in public equity markets. It seeks to invest in clean technology, software, and consumer products and services sectors. The firm prefers to in the United States and Europe. It typically invests between $5 million to $30 million in companies with revenues between $5 million to $100 million. The firm prefers to be the lead investor and take a seat on the board of directors of its portfolio companies. It prefers to hold minority as well as majority stakes. The firm also makes socially responsible investments. Next World Capital (NWC) is based in San Francisco, California with an additional office in Brussels, Belgium. Northleaf Capital Partners Toronto, Canada northleafcapital.com Northleaf Capital Partners is a private equity and venture capital firm specializing in small and mid-cap private-equity, buyouts, special situations, and venture-capital growth investments. The firm is also a fund of funds investor and invests in buyout, venture capital, distressed, mezzanine, and secondary market funds. It prefers to invest in North America with a focus on United States and Canada, Europe, and Asia with a focus on Japan. The firm actively co-invests alongside its core partner relationships as a minority investor in both private equity and infrastructure transactions. It also acts as a fund manager and advisor to its portfolio companies. Northleaf Capital Partners was founded in November 2009 and is based in Toronto, Canada with additional offices in Menlo Park, California; and London, United Kingdom. Norwest Venture Partners Palo Alto, CA nvp.com Norwest Venture Partners is a venture capital and private equity firm specializing in incubation, seed/startup, early, mid, late venture, growth equity, later stage, and recapitalization investments in both listed and unlisted companies. It invests in medical devices, specialty pharmaceuticals, healthcare services, and therapeutic products sector. For venture investments, the firm typically invests in systems and information technology infrastructure sector with a focus on

Page 160

March 14, 2012

Industry Report

embedded systems, wireless, communications systems, enterprise systems, semiconductors and components, storage, security, cloud infrastructure, and mobile; Internet and consumer sector with a focus on enabling Internet technologies, consumer Internet, social networking, digital media, social areas, advertising including online, search, consumer services, media, mobile, gaming, and retail and ecommerce; software sector with a focus on SaaS/cloud, application software, enterprise software, and mobile; and services sector with a focus on technology enabled services, communications services, infrastructure services, and financial services. For growth equity investments, it seeks to invest in information technology, technology enabled services, data and information services, business services, financial services, financial technology, consumer, and healthcare including healthcare technology sector. For its India-centric growth equity, seed, and late stage investments, the firm seeks to invest in technology, financial services, infrastructure, consumer, industrials, education, food companies, manufacturing, media, telecommunications, retail and healthcare sector. For its India-centric early to mid-stage venture capital investments, it prefers to invest in consumer, Internet, telecommunications, financial services, and education sector. The firm prefers to make investments globally with a focus on Asia-Pacific, United States, Israel, India, and China. It seeks to invest between $10 million and $15 million in its early stage portfolio companies over a period of time and can invest as little as $1 million and as much as $30 million. For growth equity companies, the firm prefers to invest between $15 million and $50 million and up to a maximum of $75 million. For growth equity investments in India, it typically invests between $15 million and $35 million and up to $75 million in each transaction. For early and mid-stage companies in India, the firm prefers to invest between $2 million and $15 million. It seeks to be a lead investor in its portfolio companies. The firm also acts as a sole venture investor or joins with other venture capital firms, corporate partners, and founding investors. Norwest Venture Partners was founded in 1961 and is based in Palo Alto, California with additional offices in Mumbai, India; Bengaluru, India; Herzliya, Israel; and Wellesley, Massachusetts. Novak Biddle Venture Partners Bethesda, MD novakbiddle.com Novak Biddle Venture Partners is a private equity and venture capital firm specializing in seed, startup, early stage, and first round investments. The firm also considers investments in later stage companies and spinouts. It typically invests in information technology companies based in the Mid-Atlantic region of the United States. The firm makes equity investments between $100,000 and $10 million in its portfolio companies. It will also syndicate larger or later stage investment rounds through its limited partners and other firms in the venture community. The firm seeks to exit between five years to seven years. Novak Biddle Venture Partners was founded in 1997 and is based in Bethesda, Maryland. Pitango Venture Capital Herzliya, Israel pitango.com Pitango Venture Capital is a venture capital firm specializing in seed, early stage, expansion, growth capital, and late stage investments. The firm seeks to invest in companies in the communication, enterprise software and Internet infrastructure, networking and storage, wireless, cellular and mobile, healthcare and life sciences, internet and media, nanotechnology, and clean technology sectors. Within communications sector, it invests in wire line infrastructure, optical networks and components, communications service providers, data networking, and communications software. For its investments in enterprise software and Internet infrastructure sector, the firm targets enterprise applications, Internet infrastructure, e-business and e-finance applications, electronic media, and security. Within network and storage, it seeks to invest in semiconductor capital equipment, fabless manufacturing, and silicon intellectual property. For its investments in wireless and cellular and mobile, it invests in wireless and cellular technologies and cellular value-added services. In life sciences sector, the firm focuses on implantable medical devices, disposable medical devices, medical equipment, telemedicine applications, and biotech drug delivery. It seeks board representation in its portfolio companies. The firm provides services such as marketing, branding, and recruitment management to its portfolio companies. It was formerly known as Polaris Venture Capital. Pitango Venture Capital was founded in 1993 and is based in Herzliya, Israel with an additional office in San Mateo, California. With over $1.4 billion under management and a growing portfolio of innovative start-ups, Pitango is one of Israels top venture capital funds. Since 1993, Pitango has invested in more than 120 technological ventures, helping them turn their ideas into viable companies and market leaders. Companies like AudioCodes; CardGuard; Convergin, acquired by Oracle; Disc-o-Tech; Dune Networks, acquired by Broadcom; Gteko, acquired by Microsoft; Jinko Solar; Optonol, acquired by Alcon; RADLAN, acquired by Marvell; RADWARE; Retalix; Scopus; Traiana, acquired by ICAP; Ventor, acquired by Medtronic; VocalTec and many more have realized success with the backing of Pitangos team. Through its offices in Israel and Silicon Valley they bring depth and breadth of sector

Page 161

March 14, 2012

Industry Report

experience and believe in a hands-on approach to working with entrepreneurs. Pitangos investment focus is diversified and balanced across investment stage and technology sectors: they invest in early, expansion and late stage companies in the fields of Communications; Networking & Storage; Wireless, Cellular and Mobile technologies; Enterprise Software, Internet and Media; Healthcare & Life Sciences, and Clean Technologies. Redpoint Ventures Menlo Park, CA redpoint.com Redpoint Ventures is a private equity and venture capital firm specializing in incubation, seed/startup, early venture, mid venture, late venture, growth capital, buyout, middle market, recapitalizations, mezzanine/subdebt, bridge, turnaround, and first round institutional financing. It invests in technology sector with a focus on cloud computing, mobile wireless platforms and applications, consumer internet services and platforms, retail, enabling technologies for the digital home, systems and software for the enterprise infrastructure companies, website infrastructure software, Internet software and services, broadband infrastructure, semiconductor materials and devices, mobile data services and internet, interactive media, content, and advertising companies. The firm also seeks to invest in energy, clean technology, and the environment sector including alternative and renewable energy, energy storage, clean coal and water purification, and energy monitoring and management solution companies. It prefers to invest in the companies based in the United States, Israel, Brazil, and China. The firm also prefers to invest in European Developed Markets. The firm seeks to be a lead or co-lead investor in its portfolio companies. Redpoint Ventures was founded in 1999 and is headquartered at Menlo Park, California with additional offices at Los Angeles, California and Shanghai, China. RRE Ventures LLC New York, NY rre.com RRE Ventures LLC is a venture capital firm specializing in investments in early stage to mid-stage privately held companies. The firm seeks to invest in high growth information technology companies, with a focus on consumer and digital media, mobility, green technology, software and services, financial technology and infrastructure. It usually seeks to invest in enterprises that are completing product development and are beginning to offer their products or services to users with an emphasis on companies selling or servicing businesses that are experiencing growth; highly dependent on information technology; or candidates who would benefit from process re-engineering (in terms of cost savings, increased revenues or both) based on microprocessor improvements, software enhancements, and rapidly increasing bandwidth capacity. The firm prefers to invest in the New York Metro region as well as Silicon Valley and other major technology innovation centers across the United States. It seeks to initially invest between $2 million and $5 million up to a limit of $10 million and $15 million in a single company over its full lifetime. The firm prefers to take a lead or co-lead all of its investments. It seeks to have a board seat in its portfolio companies. RRE Ventures LLC was founded in 1994 and is based in New York, New York. Samsung Venture Investment Corporation Seoul, South Korea samsungventure.co.kr Samsung Venture Investment Corporation is a venture capital firm specializing in investments in start-up companies. The firm typically invests in electronics, information and communication, software, internet, life science and healthcare, movies and video sectors. Samsung Venture Investment Corporation was founded in 1999, and is based in Seoul, South Korea. Sequoia Capital Menlo Park, CA sequoiacap.com Sequoia Capital is a private equity and venture capital firm specializing in incubation, seed stage, start-up stage, early stage, and growth stage investments in private companies. It also invests in public companies. The firm seeks to invest in all sectors with a focus on services including financial services, cloud computing, healthcare, consumer, Internet, outsourcing, retail, and wireless; mobile; technology; software including application, infrastructure, data management, and wireless; systems including networking, infrastructure, security, and wireless; components including semiconductors with a focus on analog and mixed signal, networking, multimedia, and programmable and wireless components; and energy including alternative energy, conventional energy, energy efficiency, energy storage, and energy services. Within financial

Page 162

March 14, 2012

Industry Report

services, it invests in banking, brokerage, payments, and enabling technology. Within healthcare, the firm invests across diagnostic services, genetics services, lab services, patient services, product development services, and enabling technology companies. Within Internet, it invests in advertising, communications, ecommerce, games, media, search, social networking, and enabling technology companies. Within the mobile sector, the firm invests in advertising, applications, devices and enabling technology companies. Within outsourcing, it invests across business process outsourcing, hosting services, managed services, professional services and software development services. Within technology, the firm invests in semiconductor, sub-systems, systems, software, and services companies. It also invests in optical components; computer, subsystems, and communication systems; and consumer and professional services. The firm seeks to invest in companies based in the United States for early and seed stage investments. However, for growth stage investments, it does not limit its investments to any country. It also invests in China, India, Israel, and Europe. It invests between $100,000 and $1 million in seed stage, between $1 million and $10 million in early stage, and between $10 million and $100 million in growth stage. The firm prefers to be the first investor and business partner in a growth stage company. Sequoia Capital was founded in 1972 and is based in Menlo Park, California. Sequoia Capital Israel Herzliya, Israel sequoiacap.com/israel Sequoia Capital Israel is a venture capital firm specializing in seed stage, early stage, start-ups, and growth investments. The firm primarily invests in the financial services, energy, internet, mobile, and technology sectors. Within financial services, it invests in banking, brokerage, and payments. Within internet, the firm invests in advertising, communications, ecommerce, games, media, search and social networking. Within the mobile sector, it invests in advertising, applications and devices. It also invests in enabling technology companies in all the aforementioned sectors. Within technology, the firm invests in semiconductor, sub-systems, systems, software, and services companies. It prefers to invest in Israel. The firm typically invests between $100,000 and $1 million in seed companies, between $1 million and $10 million in early venture, and between $10 million and $100 million in growth investments. Sequoia Capital Israel is based in Herzliya, Israel with an additional office in Menlo Park, California. It operates as a subsidiary of Sequoia Capital. Sigma Partners Boston, MA sigmapartners.com Sigma Partners is a venture capital firm specializing in seed/startups, early stage, and mid venture companies. The firm typically invests in innovative technology including software, communications, semiconductors, storage, mobile computing, electronics, infrastructure, and Internet services sectors. It primarily invests in companies based in the United States. The firm seeks to invest between $2 million and $8 million. It is typically the first venture investor and also takes a board seat in its portfolio companies. Sigma Partners was founded in 1984 and is based in Boston, Massachusetts with additional offices in Menlo Park, California and San Ramon, California. Silver Creek Ventures Dallas, TX silvercreekfund.com Silver Creek Ventures is a venture capital firm specializing in seed, start-up, and early stage investments. The firm seeks to invest in the information technology sector with a focus on security, networking, services, data communications, and optical networking. It typically invests between $2 million and $7 million per company. Silver Creek Ventures was founded in 1989 and is based at Dallas, Texas with an additional office at Denver, Colorado. Spring Creek Partners Rockford, IL Spring Creek Partners is an investment arm of Anderson Enterprises LLC specializing in all stages of development and other growth-oriented opportunities including incubation and PIPE. The firm seeks to invest in communications, software, storage, information technology, and life sciences. The firm is an S.B.I.C. Spring Creek Partners was founded in 1968.

Page 163

March 14, 2012

Industry Report

Spring Mountain Capital, LP New York, NY springmountaincapital.com Spring Mountain Capital, LP is an employee owned fund of hedge fund sponsor. The firm primarily provides its services to pooled investment vehicles, high net worth individuals, banking or thrift institutions, charitable organizations, and corporations. It invests in alternative investment markets. The firm primarily invests in hedge funds. It also seeks to make private equity investments. The firm employs an opportunistic specialty strategy to make its investments. It employs a combination of fundamental and quantitative analysis to create its investment portfolios. The firm conducts in-house research to make its investments. Spring Mountain Capital, LP was founded in 2001. Sutter Hill Ventures Palo Alto, CA shv.com Sutter Hill Ventures is a venture capital firm specializing in investments in start-up and early-stage companies. The firm invests in the technology sector. It primarily invests in companies based in the United States. The firm invests in both public and private markets. Sutter Hill Ventures was founded in 1964 and is based in Palo Alto, California. Tenaya Capital Woodside, CA tenayacapital.com Tenaya Capital is a private equity and venture capital firm specializing in growth-oriented equity or equity related investments in privately held companies. It can consider early or later stage investments on a selective basis. The firm typically invests in mid-to-later stage companies and growth investments in mature operating companies. It can also invest in earlier stage companies with limited technology development risk. The firm generally invests in a venture-backed company's second or third institutional round. It typically makes investments in companies capable of turning innovative technology and management solutions into successful businesses. The firm primarily invests in technology including Internet and consumer, IT infrastructure, communications, semiconductors and electronics, software and systems, clean technology, and healthcare industries. It seeks to invest between $5 million and $10 million in its portfolio companies with follow-on financings bringing the total to $10 million to $15 million per company. The firm can make investments as little as $3 million and as much as $20 million and can also invest more than $30 million. It seeks to invest in firms generating revenues between $1 million and $5 million in a quarter and provides follow on investments of between $10 million and $15 million. It leads or co-leads rounds of investments and serves on the board of directors. Tenaya Capital was founded in 1995 and is based in Menlo Park, California with an additional office in Boston, Massachusetts. Thomvest Ventures, Inc. Redwood City, CA thomvest.com Thomvest Ventures, Inc. is a venture capital firm specializing in start-up, early stage, and mid venture investments. It also prefers to invest in a start-ups seed, Series A, and Series B rounds. The firm seeks to invest in software and web services, networking, and semiconductors sectors. It mostly invests in companies based on the West Coast primarily in the Silicon Valley, but it can also consider opportunities in other parts of the United States and Canada. The firm typically makes initial investments between $0.25 million and $5 million in its portfolio companies. It can also make follow-on investments. The firm invests its own capital. Thomvest Ventures, Inc. was founded in 1996 and is based in Redwood City, California with additional offices in Toronto, Canada and New Providence, Bahamas. Trinity Ventures Menlo Park CA trinityventures.com Trinity Ventures is a venture capital firm specializing in seed, early, mid, late stage, and growth capital investments. It seeks to invest in technology, consumer, clean tech, infrastructure, services, and software with a focus on energy, internet advertising, digital television, consumerization of information technology, internet retail, cellular services, advertising, cloud computing, digital media, e-commerce, enterprise 2.0, financial services, gaming, lead generation, marketplace, mobile, networking, open source, SAAS, security, systems management, UGC, vertical content, video, and wireless. The firm seeks to invest in United States. It typically invests between $2 million and $20 million. The firm seeks to hold a board

Page 164

March 14, 2012

Industry Report

position in its portfolio companies. Trinity Ventures was founded in 1986 and is headquartered in Menlo Park, California with an additional office in San Francisco, California. TransLink Capital Palo Alto, CA translinkcapital.com TransLink Capital is a venture capital firm specializing in investments in early to expansion stage startup companies. The firm prefers to invest in the mobile, communications, and digital media sectors. It seeks to invest in companies based in United States, Japan, Korea, and Taiwan. TransLink Capital is based in Palo Alto, California with an additional office in Seoul, South Korea. Valhalla Partners Vienna, VA valhallapartners.com Valhalla Partners is a private equity and venture capital firm specializing in buyouts, spin-outs, early, and mid stage investments. The firm prefers to invest in the technology sector with a focus on software, internet, digital media, IT services, storage-based businesses, mobile, and networking. Within the software sector, it seeks to invest in application and integration infrastructure, development and testing, systems management, business intelligence, analytics, security, storage, Internet traffic management, supply chain management, customer relationship management, enterprise resource planning, sourcing and procurement, collaborative commerce, and business process automation. In Internet, the firm focuses on e-commerce models leveraging direct-to-consumer and peer-to-peer transactions; online marketing; online advertising; content creation and analysis tools for web content; platforms and applications that leverage the programmable web; vertical search engines; and applications. Within digital media sector it seeks to invest in content creation; visualization and analysis tools for audio, video, image, and text; applications for storage and management of digital media; distribution and monetization of user generated content; and distribution of content from traditional media publishers. In mobile sector the firm focuses on search and advertising; consumer entertainment, including games, music, and video; content management and distribution; communication, including video sharing, SMS, MMS, and social networking; enterprise and consumer applications; productivity enhancement; and infrastructure to support them all. Within storage, it seeks to invest in data protection; data movement, classification and management; Ethernet/iSCSI; disaster recovery; grid/clustering/virtualization; wide area data/file services; and next generation storage service providers. Within infrastructure, it seeks to invest in infrastructure providers for voice, data, and video networks; next-generation data centers; enterprise IT and systems management; last mile connectivity; online commerce and advertising; and computing in the cloud. The firm seeks to take a meaningful percentage ownership in the company. The firm prefers to invest in companies based in Mid-Atlantic region of United States. It seeks to exit an investment within a period of five to seven years. Valhalla Partners was founded in 2002 and is based in Vienna, Virginia. Windward Ventures San Diego, CA windwardventures.com Windward Ventures is no longer investing. The firm is a venture capital firm specializing in early stage investments. The firm does not invest in seed stage companies. It also does not invest in airlines, apparel, broadcasting, consumer products, energy, entertainment, environmental, food, housing, industrial, manufacturing, medical diagnostics, mining, publishing, real estate, restaurants, and retail sectors. The firm seeks to invest in information technology, communication and telecommunication, electronics, broadband technologies, optoelectronics, semiconductors, systems engineering, software, Internet infrastructure, data storage, and medical devices sectors. It also seeks to invest in proprietary and innovative products that target quantifiable, rapidly growing market segments and can produce revenues of over $50 million between three years and five years. The firm invests in private companies based in Southern California from Santa Barbara to San Diego. It seeks to invest $1-5M, depending on the maturity of the company. However, the firm typically invests between $3-5M over the life of an investment with initial investments as little as $250,000. It prefers to be a lead investor in early stage companies and co-invests with other venture firms in syndicated investments. It structures its investments as convertible preferred stock and provides first and second round of financing. The firm typically takes minority stake with board representation in its portfolio companies. Windward Ventures was founded in 1997 and is based in San Diego, California with additional offices in Westlake Village, California and Thousand Oaks, California.

Page 165

March 14, 2012

Industry Report

Woodside Fund Redwood Shores, CA woodsidefund.com Woodside Fund is a venture capital firm specializing in startups, seed, early stage, and late stage investments. It typically invests in technology companies with a focus on semiconductors, networking and communications infrastructure, and applications and infrastructure software, and enterprise software companies. The firm primarily invests in companies located on West Coast of United States. It typically invests between $5 million to $10 million with an initial investment of $2 million to $5 million in companies. The firm can also co-invest. It seeks to invest for a period of three years to five years and exits its investment through mergers and acquisitions and IPO. Woodside Fund was founded in 1983 and is based at Redwood Shores, California with additional offices at Woodside, California; Pleasanton, California; and San Antonio, Texas.

Page 166

March 14, 2012

Industry Report

Appendix III: Glossary of Technical Terms (definitions sourced from Storage Networking Industry Association Dictionary, 2012 and TechTarget.com)

Amazon S3 Amazon S3 is storage for the Internet. It is designed to make web-scale computing easier for developers. Amazon S3 provides a simple web services interface that can be used to store and retrieve any amount of data, at any time, from anywhere on the web. It gives any developer access to the same highly scalable, reliable, secure, fast, inexpensive infrastructure that Amazon uses to run its own global network of web sites. The service aims to maximize benefits of scale and to pass those benefits on to developers. API [General] An interface used by an application program to request services. The term API is usually used to denote interfaces between applications and the software components that comprise the operating environment (e.g., operating system, file system, volume manager, device drivers, etc.). ATA [Storage System] A standard designed to connect hard and removable disk drives. ATA is also the official name for Integrated Drive Electronics (IDE). Azure Windows-based platform for running applications and storing data in Microsoft data centers. Big Data Unstructured and semi-structured data that a company or business entity creates. Ethernet [Network] A local area networking technology based on packetized transmissions between physical ports over a variety of electrical and optical media. Ethernet can transport any of several upper layer protocols, the most popular of which is TCP/IP. Ethernet standards are maintained by the IEEE 802.3 committee. The unqualified term Ethernet usually refers to 10 Mbps transmission on multi-point copper. Fast Ethernet is used to denote 100 Mbps transmission, also on multipoint copper facilities. Ethernet and Fast Ethernet both use CSMA/CD physical signaling. Gigabit Ethernet (abbreviated GBE) transmits at 1250 Megabaud (1 Gbit of data per second) using 8b/10b encoding with constant transmission detection. Exabyte 18 [General] Shorthand for 1,000,000,000,000,000,000 (10 ) bytes. Fibre Channel (FC) [Fibre Channel] A serial I/O interconnect capable of supporting multiple protocols, including access to open system storage (FCP), access to mainframe storage (FICON), and networking (TCP/IP). Fibre Channel supports point to point, arbitrated loop, and switched topologies with a variety of copper and optical links running at speeds from 1 Gb/s to 10 Gb/s. The committee standardizing Fibre Channel is the INCITS Fibre Channel (T11) Technical Committee. High Performance Computing (HPC) The use of parallel processing for running advanced application programs efficiently, reliably and quickly. The term applies 12 especially to systems that function above a teraflop or 10 floating-point operations per second. The term HPC is occasionally used as a synonym for supercomputing, although technically a supercomputer is a system that performs at or near the currently highest operational rate for computers. Some supercomputers work at more than a petaflop or 15 10 floating-point operations per second. Host Bus Adapter (HBA) [Computer System] An I/O adapter that connects a host computer bus to a Fibre Channel or SCSI medium. Adapter is the preferred term in Fibre Channel and SCSI contexts. The term NIC is used in networking contexts such as Ethernet and token ring.

Page 167

March 14, 2012

Industry Report

I/O [Computer System] Shorthand for input/output. I/O is the process of moving data between a computer systems main memory and an external device or interface such as a storage device, display, printer, or network connected to other computer systems. This encompasses reading, or moving data into a computer systems memory, and writing, or moving data from a computer systems memory to another location. Input/output per second (IOPS) IOPS is the standard unit of measurement for the maximum number of reads and writes to non-contiguous storage locations. Internet Small Computer Systems Interface (iSCSI) [iSCSI] A SCSI Device using an iSCSI service delivery subsystem, in other words an iSCSI-specific transport mechanism (as defined by SAM-2) for SCSI commands and responses. Network Attached Storage (NAS) 1. [Storage System] A term used to refer to storage elements that connect to a network and provide file access services to computer systems. These elements generally consist of an engine that implements the file services, and one or more devices, on which data is stored. 2. [Network] A class of systems that provide file services to host computers using file access protocols such as NFS or CIFS. Network File System (NFS) [File System] [Standards] A distributed file system and its associated network protocol, commonly implemented in UNIX systems. The IETF maintains the NFS standard. NFS clients and/or servers are available for all major platforms at this point. Peripheral Component Interconnect Express (PCIe) [Computer System] A bus for connecting interface modules to a computer system. Older variations of PCI support 32 and 64 bit parallel data transfers at 33 and 66 MHz cycle times. The newer PCIe interface supports one bit wide serial "lanes" operating at 250MB/sec or 500MB/sec. Petabyte 15 [Computer System] Shorthand for 1,000,000,000,000,000 (10 ) bytes. The SNIA uses the base 10 convention commonly found in I/O-related and scientific literature rather than the base 2 50 convention (1,125,899,906,842,624, i.e., 2 ) common in computer system and software literature. Redundant Array of Independent Disks (RAID) [Storage System] A disk array in which part of the physical storage capacity is used to store redundant information about user data stored on the remainder of the storage capacity. The redundant information enables regeneration of user data in the event that one of the array's member disks or the access path to it fails. Although it does not conform to this definition, disk striping is often referred to as RAID (RAID Level 0). Storage Area Network (SAN) 1. [Network] A network whose primary purpose is the transfer of data between computer systems and storage elements and among storage elements. A SAN consists of a communication infrastructure, which provides physical connections, and a management layer, which organizes the connections, storage elements, and computer systems so that data transfer is secure and robust. The term SAN is usually (but not necessarily) identified with block I/O services rather than file access services. 2. [Storage System] A storage system consisting of storage elements, storage devices, computer systems, and/or appliances, plus all control software, communicating over a network. The SNIA definition specifically does not identify the term SAN with Fibre Channel technology. When the term SAN is used in connection with Fibre Channel technology, use of a qualified phrase such as "Fibre Channel SAN" is encouraged. According to this definition, an Ethernet-based network

Page 168

March 14, 2012

Industry Report

whose primary purpose is to provide access to storage elements would be considered a SAN. SANs are sometimes also used for system interconnection in clusters. Serial Attached SCSI (SAS) [SCSI] A SCSI interface standard that provides for attaching HBAs and RAID controllers to both SAS and SATA disk and tape drives, as well as other SAS devices. Serial Advanced Technology Attachment (SATA) [Storage System] A version of the ATA interface that uses a serial connection architecture. Small Computer System Interface (SCSI) [SCSI] A collection of ANSI standards and proposed standards that define I/O buses primarily intended for connecting storage subsystems or devices to hosts through host bus adapters. Originally intended primarily for use with small (desktop and desk-side workstation) computers, SCSI has been extended to serve most computing needs, and is arguably the most widely implemented I/O bus in use today. Service Level Agreement (SLA) [General] An agreement between a service provider, such as an IT department, an internet services provider, or an intelligent device acting as a server, and a service consumer. A service level agreement defines parameters for measuring the service, and states quantitative values for those parameters. Solid State Drive (SSD) [Storage System] A disk drive whose storage capability is provided by solid state storage. Form factors and interfaces for solid state drives are typically the same as for traditional disk drives. Total Cost of Ownership (TCO) [General] The comprehensive cost over its lifetime of a particular capability such as data processing, storage access, file services, etc. TCO includes acquisition, environment, operations, management, service, upgrade, loss of service, and residual value. Virtualization [Computer System] The act of integrating one or more (back end) services or functions with additional (front end) functionality for the purpose of providing useful abstractions. Typically virtualization hides some of the back end complexity, or adds or integrates new functionality with existing back end services. Examples of virtualization are the aggregation of multiple instances of a service into one virtualized service, or to add security to an otherwise insecure service. Virtualization can be nested or applied to multiple layers of a system. Wide Area Network (WAN) [Network] A communications network that is geographically dispersed and that includes telecommunications links.

Page 169

March 14, 2012

Industry Report

COMPANIES MENTIONED IN THIS REPORT:

Company Advanced Micro Devices Apple Inc. Aruba Networks, Inc. Brocade Communications Systems, Inc. CommVault Systems, Inc. EMC Corporation Emulex Corp. F5 Networks Inc. Fusion-io, Inc. Harmonic Inc. Intel Corporation Juniper Networks, Inc. LSI Corp. Marvell Technology Group Ltd. Mellanox Technologies, Ltd. Micron Technology, Inc. Microsoft Corporation NetApp, Inc. NetSuite Inc. Oracle Corporation QLogic Corporation Riverbed Technology, Inc. SAP AG Seagate Technology LLC STEC, Inc. Super Micro Computer, Inc Symantec Corporation VMware, Inc. Western Digital Corporation Exchange NYSE NASDAQ NASDAQ NASDAQ NASDAQ NYSE NYSE NASDAQ NYSE NASDAQ NASDAQ NASDAQ NYSE NASDAQ NASDAQ NYSE NASDAQ NASDAQ NYSE NASDAQ NASDAQ NASDAQ FSE NYSE NASDAQ NASDAQ NASDAQ NYSE NYSE Symbol AMD AAPL ARUN BRCD CVLT EMC ELX FFIV FIO HLIT INTC JNPR LSI MRVL MLNX MU MSFT NTAP N ORCL QLGC RVBD SAP GR STX STEC SMCI SYMC VMW WDC Price $7.76 $568.10 $23.40 $5.78 $51.80 $29.62 $10.42 $132.72 $32.25 $6.00 $27.49 $21.31 $8.54 $15.48 $36.76 $6.40 $32.67 $43.51 $48.95 $30.13 $17.69 $27.37 53.67 $28.00 $9.50 $16.95 $18.18 $105.08 $40.15 Rating Hold Buy Hold Buy Buy Buy Buy Buy Buy Buy Buy Hold Buy Hold Buy Buy Hold Hold Hold Buy Buy Buy Buy Buy Hold Hold Buy Buy Buy

Important Research Disclosures

Analyst Certification
We, Rajesh Ghai, CFA and Ryan Brookman, hereby certify that all of the views expressed in this research report accurately reflect our personal views about the subject securities and issuers. We also certify that no part of our compensation was, is, or will be directly or indirectly related to the specific recommendations or views expressed in this research report. The analyst(s) responsible for preparing this report has/have received compensation based on various factors, including the firm's total revenues, a portion of which is generated by investment banking activities. The analyst(s) also receive compensation in the form of a percentage of commissions from trades made through the firm in the securities of the subject company of this report, although not for any investment banking transactions with or involving the subject company. ThinkEquity LLC makes a market in Apple Inc., Advanced Micro Devices, Aruba Networks, Inc., Brocade Communications Systems, Inc., CommVault Systems, Inc., Emulex Corp., EMC Corporation, F5 Networks Inc., Fusion-io, Inc., Harmonic Inc., Intel Corporation, Juniper Networks, Inc., LSI Corp., Mellanox Technologies, Ltd., Marvell Technology Group Ltd., Microsoft Corporation, NetSuite Inc., NetApp, Inc., Oracle Corporation, QLogic Corporation, Riverbed Technology, Inc., SAP AG, Super Micro Computer, Inc, STEC, Inc., Seagate Technology LLC, Symantec Corporation, VMware, Inc. and Western Digital Corporation securities; and/or associated persons may sell to or buy from customers on a principal basis. ThinkEquity LLC has made affirmative disclosures concerning each of the covered securities mentioned in this report, including analyst holdings (if any), rating definitions and overall ratings distributions. These disclosures can be found in the most recent complete research report for each of the respective companies. Reports are available upon request.

Page 170

March 14, 2012

Industry Report

Rating Definitions
Effective October 7, 2009, ThinkEquity LLC moved from a four-tier Buy/Accumulate/Source of Funds/Sell rating system to a three-tier Buy/ Hold/Sell system. The new ratings appear in our Distribution of Ratings, Firmwide chart. To request historical information, including previously published reports or statistical information, please call: 866-288-8206, or write to: Director of Research, ThinkEquity LLC, 600 Montgomery Street, San Francisco, California, 94111. Buy: ThinkEquity expects the stock to generate positive risk-adjusted returns of more than 10% over the next 12 months. ThinkEquity recommends initiating or increasing exposure to the stock. Hold: ThinkEquity expects the stock to generate risk-adjusted returns of +/-10% over the next 12 months. ThinkEquity believes the stock is fairly valued. Sell: ThinkEquity expects the stock to generate negative risk-adjusted returns of more than 10% during the next 12 months. ThinkEquity recommends decreasing exposure to the stock. Distribution of Ratings, Firmwide ThinkEquity LLC
IB Serv./Past 12 Mos.
Rating Count Percent Count Percent

BUY [B] HOLD [H] SELL [S]

125 52 11

66.50 27.70 5.90

13 1 0

10.40 1.92 0.00

This report does not purport to be a complete statement of all material facts related to any company, industry, or security mentioned. The information provided, while not guaranteed as to accuracy or completeness, has been obtained from sources believed to be reliable. The opinions expressed reflect our judgment at this time and are subject to change without notice and may or may not be updated. Past performance should not be taken as an indication or guarantee of future performance, and no representation or warranty, express or implied, is made regarding future performance. This notice shall not constitute an offer to sell or the solicitation of an offer to buy, nor shall there be any sale of these securities in any state in which said offer, solicitation, or sale would be unlawful prior to registration or qualification under the securities laws of any such state. This research report was originally prepared and distributed to institutional clients of ThinkEquity LLC. Recipients who are not market professionals or institutional clients of ThinkEquity LLC should seek the advice of their personal financial advisors before making any investment decisions based on this report. Additional information on the securities referenced is available upon request. In the event that this is a compendium report (covers more than six ThinkEquity LLC-covered subject companies), ThinkEquity LLC may choose to provide specific disclosures for the subject companies by reference. To request more information regarding these disclosures, please call: 866-288-8206, or write to: Director of Research, ThinkEquity LLC, 600 Montgomery Street, San Francisco, California, 94111. Stocks mentioned in this report are not covered by ThinkEquity LLC unless otherwise mentioned. Member of FINRA and SIPC. Copyright 2012 ThinkEquity LLC, A Panmure Gordon Company

Page 171