The document provides an overview of networking issues and infrastructure related to cloud computing and data centers, including classifications, performance challenges, and the role of TCP/IP. It discusses the architecture of data centers, the nature of traffic, and the importance of high availability and security in cloud services. Additionally, it highlights the benefits and drawbacks of different networking designs and mentions key cloud service providers and their offerings.
The document provides an overview of networking issues and infrastructure related to cloud computing and data centers, including classifications, performance challenges, and the role of TCP/IP. It discusses the architecture of data centers, the nature of traffic, and the importance of high availability and security in cloud services. Additionally, it highlights the benefits and drawbacks of different networking designs and mentions key cloud service providers and their offerings.
The document provides an overview of networking issues and infrastructure related to cloud computing and data centers, including classifications, performance challenges, and the role of TCP/IP. It discusses the architecture of data centers, the nature of traffic, and the importance of high availability and security in cloud services. Additionally, it highlights the benefits and drawbacks of different networking designs and mentions key cloud service providers and their offerings.
The document provides an overview of networking issues and infrastructure related to cloud computing and data centers, including classifications, performance challenges, and the role of TCP/IP. It discusses the architecture of data centers, the nature of traffic, and the importance of high availability and security in cloud services. Additionally, it highlights the benefits and drawbacks of different networking designs and mentions key cloud service providers and their offerings.
► Understand the general classification of data centers
► Present an overview of the data center environment ► Understand the basic networking issues in data centers ► Explain the performance challenges faced by TCP/IP in data center networks ► Describe the newly designed TCPs for data center networks and their novelty NETWORKING FOR CLOUD COMPUTING - INTRODUCTION ► Routers , Switches, Cables, WAN & LAN Connections ► OSI TCP/IP CONCEPTUAL VIEW A LAN LOGICAL VIEW OF WAN MPLS LABEL DISTRIBUTION PROTOCAL GLOBAL MPLS VLAN ► VLAN single broadcast domain, that appears to be working on the same LAN, Even they are in different geographical locations. ► It is not restricted to a physical boundaries in a switched Network. ► It can be spread across multiple switches in a Network or even managed in a single switch. ► Vlan ranges from 0 to 4095. ► 2 Types of Ranges Normal Range Vlan (1 to 1005) ► Extended Range Vlan (1006 to 4095) VLAN VLAN DATA CENTER OVERVIEW DATA CENTER OVERVIEW BLADE SERVER COST OF THE SERVER Overview of Data Center Environment 1. Architecture of Classical Data Centers ► Multitier model – HTTP, ► The multitier approach includes web, application, database layered design supporting commerce and enterprise business ERP and CRM solutions ► Multitier model uses software that runs as separate processes on the same machine using inter process communication (IPC). ► Server cluster model has grown out of the university and scientific community to emerge across enterprise business verticals including financial, ► manufacturing, and entertainment ► The following three tiers are used 1. Web server 2. Application 3. Database Overview of Data Center Environment 2. CEDCs New technological and economic pressures have forced organizations to look at ways to get more out of their IT infrastructure. The cloud has the capability to reduce costs and increase the flexibility of applications and services including IT infrastructure. The virtualized environment is transformed into one that is optimized with intelligent, integrated IaaS, and PaaS that manages dynamic workload giving the workloads the resources that it needs based on business policies. Overview of Data Center Environment 3.Physical Organization
Data centers generally span across the entire building,
a few floors of a building or even a single room in the building The several possible ways of arranging the server racks in a data center. Data centers usually comprise of a large number of servers that are mounted in rack cabinets and are placed in single rows forming corridors (so-called aisles) between them. Large data centers house several thousand servers wherein sometimes shipping containers packed with 1000 or more servers each are used. Overview of Data Center Environment 4. Storage and Networking Infrastructure 1. Client–server network 2. Server–server network 3. Server–storage network 4. Other networks
A two-tier network topology is very popular in DCNs today. Access switches for server connectivity are collapsed in high-density aggregation switches that provide the switching and routing functionalities for access switching interconnections and various LAN servers. Overview of Data Center Environment BENEFITS Storage and Networking Infrastructure 1. Design simplicity (fewer switches and so fewer managed nodes) 2. Reduced network latency (by reducing the number of switch hops) 3. Typically a reduced network design oversubscription ratio 4. Lower aggregate power consumption
Disadvantage of a two-tier design includes limited scalability:
when the ports on an aggregation switch pair are fully utilized, then the addition of another aggregation switch/router pair adds a high degree of complexity. Overview of Data Center Environment 5. Cooling Infrastructure data centers usually span across the entire building and house several thousand servers, a sophisticated cooling infrastructure is deployed, which may involve building level air-conditioning units, fans, and air recirculation systems.
Green technologies and environmental awareness have also been a large part of the industry in the past 3 years. No longer just a choice between build and lease, the data center can be owned, placed in colocation, wholesale, put in a public or private cloud, or a hybrid strategy of options. Overview of Data Center Environment 6. Nature of Traffic in Data Centers Data center environment is largely different from that of the Internet, for example, the round-trip time (RTT) in DCNs can be as less as 250 μs in the absence of queuing. The reason is that DCNs are privately owned networks tailored to achieve high bandwidth and low latency.
Mice traffic : Google search and Facebook updates
Cat traffic : Small- and Medium-sized file downloads
Elephant traffic : Antivirus updates and movie downloads Networking Issues in Data Centers Cloud networking is the fundamental backbone to provide cloud services and the reason behind the shift in how IT services are provided to users.
Availability
cloud provider organization faces is to provide maximum uptime for the services that are offered to the users.
Few seconds of downtime may lead to loss of reputation for the organization and affect the overall business. downtime may lead to violation of service-level agreements (SLAs) between the cloud user and the cloud provider, thus largely affecting the cloud provider’s revenues.
high availability is to replicate the data and take regular backups.
Networking Issues in Data Centers Poor Network Performance Three basic performance requirements of a DCN are high burst tolerance, low latency, and high throughput. data center comprises a mix of all the three kinds of traffic: mice traffic, cat traffic, and elephant traffic, each having a different performance requirement major challenge is that traditional Transmission Control Protocol/ Internet Protocol (TCP/IP) stack, which is mainly designed for Internet-like scenarios, fails to provide optimal performance in DCNs
Security cloud user’s data secure during transit, or while it is at rest, is yet a concern for the cloud service providers. Accidental deletion of data because of a power outage or malfunctioning of a regular backup program may lead to loss of customer’s data and incur huge damage for the hosting organization. TRANSPORT LAYER ISSUES IN DCNS ► TCP is one of the most dominant transport protocols, widely used by a large variety of Internet applications, and also constitutes majority of the traffic in both types of DCNs. ► TCP can be partly attributed to the congestion control mechanisms implemented. ► TCP fails to satisfy the three basic requirements together within the time boundaries ► TCP incast, ► TCP outcast, ► Queue buildup, ► Buffer pressure and pseudo congestion effect TRANSPORT LAYER ISSUES IN DCNS CONGESTION CONTROL CONGESTION Avoidance, Detection CONGESTION CONTROL TCP INCAST ► High “Fan – In” ► High bandwidth, low latency ► Buffers in the switches is quite small, lots of parallel requests each small amount of data ► Bursty retransmissions ► TCP timeout – RTT timout. TCP OUTCAST ► TCP outcast and mainly occurs in data center switches that employ drop-tail queues ► Drop-tail queues lead to consecutive packet drops from one port and, hence, cause frequent TCP timeouts of drop-tail queues is termed as port blackout QUEUE BUILDUP ► mice traffic, cat traffic and elephant traffic coexist in a DCN. ► when both mice traffic and elephant traffic traverse through the same route, the performance of mice traffic is significantly affected due to the presence of the elephant traffic BUFFER PRESSURE ► long-lasting and greedy nature of elephant traffic. ► large number of packets from mice traffic are lost, leading to poor performance. ► mice traffic get dropped frequently because the elephant traffic lasts for a longer time and keeps most of the buffer space occupied PSEUDOCONGESTION EFFECT ► Modern data centers adopt virtual machines (VMs)to offer on-demand cloud services and remote access ► virtualized data centers ► virtualization like efficient server utilization, service isolation, and low system maintenance cost ► Hypervisor scheduling latency ► RTO occurs, VM sender assumes that the network is heavily congested and significantly brings down the sending rate. Cloud Service Providers ► Amazon, Microsoft, Google, Yahoo, EMC, Salesforce, Oracle, IBM. ► EMC IT - virtualizing the infrastructure, allocation of the resources on demand is possible. EMC IT provides its business process units with IaaS, PaaS, and SaaS. ► IAAS – individually or integrated services ► PAAS - database platforms ► SAAS – ERP, CRM CAPTIVA CLOUD TOOLKIT - EMC ► Scan - Scanning is importing activity of documents pdf, tiff and jpg. ► MDW - Multi Directory Watch is another entry point to Captiva, Attached file in an e-mail. MDW also acts as a scan module except it does not interlock with the scanner. ► IE - Image enhancement is a kind of filter or repairing tool for images that are not clear. ► Index - Indexing is a data capturing activity in Captiva through which one can capture key data from various fields. ► Export - Export is the exit point of Captiva where images/data are sent to various repositories like file, net, document, or data. ► Multi - Multi is the last process in Captiva to delete batches that have gone through all modules and exported value successfully. GOOGLE ► Google is one among the leading cloud providers that offer secure storage of user’s data. ► cloud platform, app engine, cloud print, cloud connect, and many more features that are scalable, reliable, as well as secure. Cloud Platform Google Cloud Platform enables developers to build, test, and deploy applications on Google’s highly scalable and reliable infrastructure. Software infrastructures such as MapReduce, BigTable, and Dremel are the innovations for industrial development. Google Cloud Platform includes virtual machines, block storage, NoSQL datastore, and big data analytics. It provides a range of storage services that allow easy maintenance and quick access of user’s data. GOOGLE Cloud Storage Google Cloud Storage is a RESTful online file storage web service for storing and accessing one’s data on Google’s infrastructure. Google Cloud Connect Google Cloud Print Google App Engine
