A Study Report on Software Defined Data Center

Ashok Kumar Chaudhary (I201921273)

Huazhong University of Science and Technology
School of Electronic Information and Communication
Wuhan, China
[email protected]
Abstract

Data center is a network infrastructure for carrying, transmitting, storing and

processing huge data, which plays an important role for the fast growth of cloud
computing used in our daily life. The rapid growth and the distributed locations of the
datacenters increase the complexity of control and management processes. A new
paradigm which is called Software Defined Networking (SDN) comes as a solution to
reduce the overhead of datacenters management by abstracting all the control
functionalities from the hardware devices and setting it inside a software layer. These
functionalities are responsible for organizing, controlling and managing the main
components of the data center; network, storage, compute and security. The Software
Defined Datacenter (SDDC) integrates the software defined concepts into all of these
components.

This paper presents an over view of Software Defined Data Center (SDDC) architecture
and benefit with the integration of SDN in Data Center Network.

Keywords: Data Center Network (DCN), Software Defined Network (SDN), Software
Defined Environment (SDE), Software Defined Data Center (SDDC), Software Defined
Compute (SDCompute), Software Defined Storage (SDStore), Software Defined Security
(SDSec), Virtualization and Network Function Virtualization (NFV).

I. INTRODUCTION analysis, and storage are increasing. To meet

the service demands effectively, data centers
Recently, the rapid development of large
should always provide high quality services.
data centers is promoted by the ever
Traditionally, computation and storage are
increasing computing resource demand. Data
spread across hundreds and thousands of
centers are deployed worldwide and the
servers in a data center network (DCN) as a
distributed services running in modern data
centers such as massive data processing,
1
distributed way so that achieving a stable The advantages that SDN offers to network
communication solution is very important. management come to fill these gaps. By
decoupling the control from the data plane,
One of the most important requirements for
forwarding devices become much simpler
data center networks is to find ways to scale
and therefore cheaper. At the same time all
in order to support hundreds of thousands of
control logic is delegated to one logically
servers and millions of virtual machines.
centralized entity. This allows the dynamic
However, achieving such scalability can be a
management of flows, the load balancing of
challenging task from a network perspective.
traffic and the allocation of resources in a
First of all, the size of forwarding tables
manner that best adjusts the operation of the
increases along with the number of servers,
data center to the needs of running
leading to a requirement for more
applications, which in turn leads to increased
sophisticated and expensive forwarding
performance. Finally, placing middle boxes
devices. Moreover, traffic management and
in the network is no longer required, since
policy enforcement can become very
policy enforcement can now be achieved
important and critical issues, since
through the controller entity.
datacenters are expected to continuously
achieve high levels of performance. All infrastructures in an SDDC can be
delivered as a service leveraged as a private,
In traditional datacenters the aforementioned
hybrid, or public cloud. SDDC substantially
requirements are typically met through the
reduces CapEx and Opex, allowing
careful design and configuration of the
deployment of applications in minutes with
underlying network. This operation is in
automated management system. It achieves
most cases performed manually by defining
new levels of infrastructure utilization and
the preferred routes for traffic and by placing
employee productivity.
middle boxes at strategic choke points on the
physical network. Obviously, this approach Software Defined Networking (SDN) is a
contradicts the requirement for scalability, very recent paradigm, proposed to address
since manual configuration can become a control and management challenges known
very challenging and error-prone task, in traditional platforms by hiding their
especially as the size of the network grows. complexities from the end users. This is
Additionally, it becomes increasingly achieved by isolating the data plane from the
difficult to make the data center operate at its control plane. SDN technology has grown
full capacity, since it cannot dynamically very rapidly to encompass a number of
adapt to the application requirements. disciplines such as Networking (SDN),

2
Storage (SDStore), Security (SDSec),
Compute (SDCompute), Data Centers
(SDDC), etc.

II. Methodology

Methodology applied by the study was

“Desktop Review”. The study consulted
variety of internet sources to highlight
evidenced facts on this study review. Figure 1 Software Defined Data Center
Possible data search used was website and
SDDC is defined as to virtualize all
some journals where convincible and
infrastructure components in a data center
accessible in web territories to adhere the
that include compute, storage, network, and
conclusion.
security with comprehensive abstraction,

III. Software Defined Data Center pooling, and automation by software with
little or no human involvement. SDDC is a
(SDDC)
new paradigm which simplifies the control
In a traditional data center, the infrastructure and management of the system. This concept
typically consists of hardware and devices is a comprehensive component to different
that could take weeks to install. With software defined systems which are the
increasing big data from mobile, social Software Defined Compute (SDCompute),
media, public, and internet of things, they Software Defined Network (SDN), Software
demand heavier applications for analytics, Defined Storage (SDStore), and Software
storage, and network. IT needs to move Defined Security (SDSec). SDDC provides a
much faster to deploy increasingly larger and solution to support both legacy enterprise
complex infrastructure from weeks to days applications and cloud computing service.
or minutes. An SDDC, pioneered by Figure 2 shows these main components of
VMware and recognized by industry, the SDDC.
provides a faster, smarter, and inexpensive
solution to significantly improve IT
efficiency and performance.

3
 workloads. Server virtualization is
Application Layer (User Interface)
considered as the father of the SDDC,
Orchestration Layer (Data Center OS)
because compute was the first to mature and
Software Based Management/Control Layer garner mainstream adoption. The IT industry
SDCompute SDStorage SDN SDSecurity as a whole is already very familiar with this
practice, and it is widely accepted as the
Physical Layer (Infrastructures of Data Center) standard for deploying generic, mixed-
Compute Storage Network Security
workload server infrastructures.
System
Figure 2 The Architecture of SDDC and its main
The virtualization is used to accelerate the
components.
run time and let more than one simulation,
Software Defined Compute (SDCompute): user, and task to work at the same time by
Software Defined Compute (SDCompute) encapsulating and isolating each one on a
means virtualization of the data center’s separate VM and let many VMs to work
server and compute segment. Server simultaneously. In such a way, the system
Virtualization means abstraction of compute resources will be exploited very well and
resources (CPU and memory) that can be critical decisions will be taken faster.
realized with underlying collection of Moreover, the system can scale up to cover
physical server resources. As a standard many locations and users.
infrastructure technology, server
The challenge for other SDDC resources like
virtualization is the basis of the SDDC,
storage and networking is to get adoption
which extends the same principles to all
and trust of the technologies to become main
infrastructure services. One of the main
stream. SDCompute has already reached that
advantages of SDCompute is that it enables
point, so the challenge for it today is to add
the infrastructure to become more flexible,
more control and intelligence to the
serving many clients or customers at the
platform. Software defined is all about the
same time.
ability to manipulate the system from outside
In a post-virtualization data center, CPU and the system typically via Application
memory are grouped together as “compute,” Programming Interfaces, or APIs.
and networking and storage are handled
Software Defined Network (SDN): It
separately. SDCompute, then, is the practice
wouldn’t be a data center without
of controlling and automating abstracted
networking! Networking is the backbone of
compute resources. In most cases, that would
everything that happens within and outside
mean manipulating virtual machine

4
the data center. Because connectivity it so
integral to the function of a data center,
abstracting network functions and topologies
into software will substantially change the
way that data centers are designed and
operated. Network abstraction also unlocks
new levels of scale and flexibility with
technologies.

SDN is the latest innovation in network

environment. As any software defined
system, the SDN simplifies the network
management by decoupling the control plane
Figure 3 General SDN Architecture
from the data plane and allows
In network communication, the messages
programmatic interaction with the control
created by the user present the data plane;
plane. This change allows the network to be
these messages need to be transferred to an
redesigned on the fly or scaled to meet
appropriate destination. Network
demand. Traditional networking has relied
management is responsible for finding the
heavily on link oversubscription, and as
best path to send the message by using the
virtualization has changed the game, SDN
control messages while maintaining any
allows network architects to create dynamic
information related to network like network
networks that can be adapted to meet
traffic. In SDN the data plane and the control
changing requirements.
plane are separate, where the data plane uses
SDN removes the dependence on specific the forwarding tables prepared by the control
manufacturers and vendors as the physical plane in the controller to forward the
hardware is abstracted. This means that IT is messages, flow-packets.
no longer bound by the product life cycle
Software Defined Storage (SDStore): The
and development iterations of a given
idea of SDStore is similar to the Software
vendor. Rather, they can move hardware in
Defined Networking (SDN) approach in
and out at will underneath their abstraction
separating the control plane from the data
layer. This is made possible by projects like
plane. In SDN, the central controller handles
OpenFlow, which has become almost
all the traffic in the network by defining the
synonymous with SDN.
forwarding rules and pushing them to the

5
switches to forward the packets to their 4 illustrates the architecture of the SDStore
destinations. system and shows its key components in
general.

The architecture of SDStore consists of three

layers: infrastructure layer, control layer, and
application layer. The Infrastructure layer
combines various storage devices storing the
raw data. The controller in the control layer
is considered the most critical element in
SDStore systems. It is where the storage
resources in the infrastructure layer interact

Figure 4 The SDStore Architecture with the application layer. The control layer
SDStore is one of the most important converts different policies to different
subsystems in SDN. It takes the instructions inside the system. The last layer
responsibility of managing huge data in in this architecture is the application layer
storage systems by isolating the data control which holds different applications and
layer from the data storage layer. The control allows the end user to interact with storage
layer refers to the software component that devices.
manages and controls the storage resources,
Software Defined Security (SDSec):
whereas the data layer refers to the
Traditional security mechanisms are not
underlying infrastructure of the storage
suitable for the new SDN architectures. The
assets. Such isolation will reduce the
design of traditional security devices is
management complexity with this new
unable to protect the components of
system architecture design. Moreover, it will
virtualized environments, due to its
reduce the cost of the infrastructure by
dependency on physical network devices,
creating a single central control unit to
which cannot see the significant security
manage the different elements in the system
activities inside virtualized environments.
regardless of their vendors rather than
The changes brought by virtualization,
installing the control software on each
which range from new virtual network
element. The controller (software layer)
topology and the threats related to the
applies various policies on different types of
hypervisors to eliminating the roles in
storage elements according to data flows,
virtualized management, demonstrate the
i.e., requests by different application. Figure
need of new virtualized security

6
mechanisms. Such virtualization would and devices. On the other hand, the software
reduce the complexity and the cost of deployed concept means that the
security operations. Software Defined functionality of the service is deployed in a
Security (SDSec) is a new technology computer hardware object. Using software to
emerging under the SDN paradigm. It is an manage and control the resources is not a
example of a Network Function new concept. The essential difference, which
Virtualization (NFV). The new technology was brought by software defined concept, is
provides a new way to design, deploy and the ability of the control layer to control all
manage security mechanisms by separating the underlying resources regardless of their
the forwarding and processing plane from vendor variations by physically isolating
the security control plane, in a similar way them from the hardware resources in the data
as SDN abstracts the forwarding plane from layer.
the control and management plane. Such
separation provides a scalable distributed
security solution, which virtualizes the
security functions but remains manageable
as a single logical system. SDSec is a
solution to help secure virtualized
environment infrastructures, including
virtual network, virtual storage and even Figure 5 The workflow and policy enforcement in SDSec
virtual servers from different threats whether
Similar to the SDN architecture, the
they are traditional such as intrusion
architecture of SDSec separates the data
detection and denial of service attacks or
plane from the control plane and consists of
specific to virtualized environments such as
three main layers-the physical layer, the
insider threats. In SDSec the functions of
control layer and the application layer.
network devices, like intrusion detection,
Inside the Physical layer, all the hardware
firewalling and others, are extracted from the
devices are located, which may include
hardware appliances to a software layer.
database arrays, switches, routers, or any
It is important to differentiate between the other asset. The Physical layer is also called
software defined concept and another related the data layer and its role is limited to
concept which is “software deployed.” In the following the policies created by the control
former concept, the APIs and software are layer, which resides on top of this layer. All
used to control and manage the resources the control and management operations are

7
abstracted form the devices located in the
up facility in ways that would have been
physical layer and set inside the Control
unthinkable at the turn of the century. By
layer (Middleware layer). The Control layer
greatly expanding their capacity, they’ve
is considered the brain of any SDN since it
managed to offer a wider range of services in
handles all the core control operations. The
Application layer is the only one visible to a relatively cost-effective manner. Building a

the user. All applications are implemented new data center remains a massive upfront

inside the Application layer. Several investment, but SDDC allow those facilities
applications can be created on the top of the to provide more value per unit area than ever
control layer to help the users interact with before, making large scale data centers not
the underling devices and data. only possible, but practical as well.

The three layers can communicate with each Accessibility: With no physical hardware to
other by a set of APIs, southbound and manipulate or connect to the SDDC,
northbound. The former is used by the customers can access services and
physical layer to interact with the control proprietary data stored in a virtualized
layer. The most famous example of such environment from anywhere at any time. For
APIs is the OpenFlow protocol, which is companies pushing the boundaries of edge
used in SDN systems. As for the latter, it is computing with IoT devices and streaming
used by the user’s applications to interact content services, server virtualization
with the control layer. benefits have fundamentally changed the
way they design and implement their
IV. Benefits of SDDC
network architecture. SDDC can manage
In a Software Defined Data Center (SDDC), customer resources faster and more

policy-driven automation enables effectively and also allow SDDC customers

to monitor and manage their own virtualized
provisioning and ongoing management of
environments.
logical compute, storage and network
Scalability: One of the biggest advantages
services. The result is unprecedented IT
SDN offer SDDC customers is its ability to
agility and efficiency, with flexibility to
rapidly scale to meet their particular needs.
support today and tomorrow's hardware and
With additional storage capacity and
applications. Benefits include: computing power readily accessible through
Cost Savings: In SDDC, SDStorage and software defined environment, organizations
SDCompute have made it possible to scale experiencing rapid growth or increased

8
demand can expand their capacity without measures to big data analytics driven by
having to purchase additional hardware. machine learning, the modern data center is
Data center as a service (DCaaS) platform a virtual marketplace that offers companies a
model provides unmatched flexibility for wide range of services that can benefit their
companies with variable workloads. business as they grow over time. Thanks to
SDN paradigm, these organizations only pay
Efficiency: Energy demands of Data Center
for the services they receive, which allows
is very high i.e., about three percent of the
them to devote their resources to offerings
entire world’s electricity. Given all the
that add real strategic value.
power they use, it’s hard to believe that these
facilities are, on average, far more efficient V. Conclusion
than their predecessors. SDN deserve a lot of
The Software-Defined Data Center (SDDC),
credit for this shift by moving from a purely
a tested architectural approach based on the
hardware systems to software-driven server
concept of virtualization and automation,
management, SDDC has become much more
drives many of now day's leading data
efficient and have been able to do far more
centers. The SDN approach enables ICT
with far less. Innovations in memory,
industries to adopt SDDC technologies at
processors and cooling have been the biggest
their own pace, without having to rip and
contributors to server virtualization benefits,
replace the existing infrastructure. The
greatly expanding the capabilities of the
SDDC concept changes both the physical
hardware that makes up a data center’s
infrastructure and the processes of the data
physical infrastructure. High-density rack
center and thus influence the ICT strategy.
deployments are just one example of how
The SDDC enables ICT sector to adopt a
computing resources have become much
hybrid cloud strategy and empowers
more efficient in recent years. SDDC
enterprises to achieve outcomes that enhance
architecture has redefined the nature and role
efficiency and security. The importance of
of today’s data centers.
SDDC is highlighted as it provides a
And it’s not just about getting more of what centralized, programmable, flexible, simple
they’re already using. SDN paradigm allows and scalable solution to control and protect
SDDC to offer a wide range of services that the systems with the rapid increase of
would have been unthinkable in past datacenter
decades. From enhanced security

9
Acknowledgment

I would like to thank the School of Electronics Information & Communication of Huazhong
University of Science & Technology, especially Prof. Jingwen CHEN, for delivering the
knowledge of new network technologies in my study as well as directions to write a technical
study report paper. I also express my thanks to friends and country mate for their suggestions
about report preparation that improves my writing skill during the work.

References
[1] “Cisco Global Cloud Index: Forecast and Methodology, 2016–2021,” White Paper,
[Online]Available:http://www.cisco.com/c/en/us/solutions/collateral/serviceprovider/glob
al-cloud-index-gci/Cloud_Index _White_Paper.pdf
[2] D. Shrimankar and M. Paliwal "Effective Resource Management in SDN enabled Data
Center Network based on Traffic Demand" IEEE Access Vol. 4, 2016
[3] W Xia, Y. Wen, C. H. Foh, D. Niyato, and H. Xie "A Survey on Software-Defined
Networking" IEEE Commun. Surveys Tuts, vol.17, no.1, pp.27-51, 2015.
[4] R. de Oliveira, A. Shinoda, C. Schweitzer, and L. Rodrigues Prete, “Using mininet for
emulation and prototyping software-defined networks,” in Communications and
Computing (COLCOM), 2014 IEEE Colombian Conference on, June 2014, pp. 1–6.
[5] K. Palanivel and B. Li, “Anatomy of software defined storage challenges and new
solutions to handle metadata,” University of Minnesota, Report, Aug 2013.
[6] Jagadeesh CV "Whitepaper: Software Defined Data Center–An Implementation view"
Happiest Minds Report, 2016.
[7] “The fundamentals of software-defined storage” simplicity at scale for cloud
architectures”,” Coraid Inc, Technical Report, 2013.
[8] C. Hollis, “Introducing emc vipr: A breathtaking approach to software defined storage,”
http://chucksblog.emc.com/chucks blog/2013/05 / introducing-emcvipr-a-breathtaking-
approach-to-software-defined-storage.html [Online; accessed Oct-2014], 2013.
[9] Y. C. Huang, Y. Yoshida, K.I. Kitayama, S. Ibrahim, R. Takahashi and A. Hiramatsu,
“Modeling and performance analysis of OPS data center network with flow management
using Express Path,” 2014 International Conference on Optical Network Design and
Modeling, Stockholm, 2014, pp. 90-95.
[10] R. Miller, “Who Has the Most Web Servers?, 2013. [Online].http:// www.
datacenterknowledge.com/archives/2009/05/14/whosgotthe-most-web-servers/
[11] J. R. Hamilton, N. Jain, S. Kandula, C. Kim, P. Lahiri, D.A.Maltz, P. Patel, and S.
Sengupta, “VL2: A scalable and flexible datacenter network,” Proc. ACM SIGCOMM
Conf. Data Commun., pp. 51-62, Aug. 2009.
[12] Ningning Hu and P. Steenkiste, “Improving TCP startup performance using active
measurements: algorithm and evaluation,” 11th IEEE Int. Conf. on Network Protocols,
2003, Proceedings, pp. 107-118.
[13] Open Networking Fundation. Software-defined networking: The new norm for
networks. ONF White Paper, 2012.
10
[14] W. Xia, et al, "A Survey on Data Center Networking (DCN): Infrastructure and
Operations," IEEE Commun Surveys & Tutorials, 2017, vol.19, no.1, pp.640-656
[15] M. Al-Fares, et al, "A Scalable, Commodity Data Center Network Architecture,"
Proc. SIGCOMM'08, Aug. 2008
[16] C. Kim, et al, "Floodless in SEATTLE: A Scalable Ethernet Architecture for Large
Enterprises," Proc. SIGCOMM'08, Aug. 2008
[17] N. McKeown, T. Anderson, H. Balakrishnan, G. Parulkar, L. Peterson, J. Rexford, S.
Shenker and J.Turner "OpenFlow: Enabling Innovation in Campus Networks," ACM
SIGCOMM Computer Communication Review, vol. 38, no. 2, 2008, pp.69-74
[18] D. Kreutz, F.M. V. Ramos, P. E. Verissimo, C. E. Rothenberg, S. Azodolmolky and
S. Uhlig, "Software-Defined Networking: A Comprehensive Survey," Proceedings of the
IEEE, vol. 103, no. 1, 2015, pp.14-76
[19] “Software defined data center (SDDC) definition,” DMTF, white paper, 2014.
[20] R. Jain and S. Paul, “Network virtualization and software defined networking for
cloud computing: a survey,” Commun. Magazine, IEEE, vol. 51, no. 11, pp. 24–31,
November 2013.

11