DCNSS Exam - Cisco Data Center 3.0

 

Next Gen Data Center

Cisco has identified three networking environments in which the intelligent network can play a vital role. These are data networking, storage networking, and application networking. These represent three functional areas which in some cases overlap.

The traditional data network infrastructure, the LAN and the WAN, are well established with IP protocols being ubiquitous. The storage area network (SAN) is increasing in the data center, leveraging both Fibre Channel and IP protocols. Application networking is a less well known term in that it refers to embedding specific application platforms in a network that take on low-level, repeatable, and unchanging functions that remove a burden from the application running in the server. An example of this may be security where the requirement goes beyond the typical firewall deliverables.

Cisco’s Intelligent Information Network (IIN) strategy defines a three-to-five year roadmap for increasing value through embedding greater intelligence in a ubiquitous network. Cisco’s Service-Oriented Network Architecture (SONA) provides the roadmap to build an SOI using Cisco’s Data Center Network Infrastructure (DCNI) products in three main phases: Consolidation, Virtualization, and Automation. These three phases will run in parallel and will all evolve at different speeds for distinctive technology areas and are part of Cisco’s Service Oriented Data Center (SODC) initiative.

Within many data centers, the responsibility for different technology and functional areas is split. You may find different teams dealing with servers, storage, and the data network.

This lack of a holistic approach to the data center can mean that each technology environment deals with scalability, virtualization, management, and security without reference to the other areas in the data center. This can lead to inefficient use of both capital expenditure and expensive data center staff as well as limiting the ability to deploy and redeploy resources, flexibly, across the data center.

Cisco’s strategy of having a converged, intelligent network as a foundation for the data center facilitates taking a holistic, cross environment, view of the data center using network-embedded functions that can be effective across all the data center environment. This can mean reduced OpEx for the data center as the network offers a single point of management for the whole data center. Reducing the number of functions duplicated within different environments can also mean improvements in data center CapEx. As functions become embedded in the network there is also the added value of having highly resilient, director-class network devices which can not only consolidate the functions, but also result in improved service availability.

This concept is an important element of maximizing the sales potential in the data center.

Application networking is a term used to describe network-embedded applications which are shared or networked between different devices and applications. This provides for a single instance of the function within the network to be made available, end to end, across the data center.

Cisco has two main categories of Application Networking Solutions. Data center solutions provide secure, reliable, high-performing data center infrastructure for application traffic within or across data centers. Branch office and WAN solutions provide LAN-like performance for applications delivered over the WAN, and thus enable IT to consolidate servers/storage within the branch office. Cisco branch solutions for application delivery are deployed symmetrically in both the data center and branch offices.

Data Center Challenge

Today’s enterprise networks face common challenges and problems across the board when looking to deploy application solutions. The lowest common denominator for each category gives us four distinct challenges. The first of these is availability and reliability. Is the application available when needed? Is it reliable or do I have to constantly monitor it and “nurse” it? The second challenge is performance. Does the application perform well or are the end users spending too much time at the water cooler? Security is always a challenge. Who can access the data? The last challenge is cost. How much does it cost to deploy, operate, and maintain the application?

 

Data Center 3.0

When you consider virtualization across the data center, you always have to start with a physical device or resource. These include servers, switches, and storage. These physical devices and resources are dependent upon a foundation network to communicate. It is this network that Cisco is unifying into a single fabric based on Ethernet, which is a core element of Cisco’s Data Center 3.0 strategy. Within this unified fabric, Cisco has embedded its network services.

In order to virtualize the data center, it is necessary to consolidate the data center resources and then deploy them as logical devices and resources. Server virtualization is accomplished through techniques such as VMWare, Windows Virtual Server, and Virtual Iron, each with their own features, functions, benefits, and price points. Virtualizing the data network can be handled in more than one way. Typically, VLANs allow the network to be logically segmented, but with the advent of the Nexus 7000, Cisco can now also deploy virtual domains of switch functionality—effectively virtual switches—also known as virtual device contexts, or VDCs. The Nexus 7000 supports a maximum of 4 VDCs in NX-OS release 4.1.

In the storage networking environment, Cisco’s MDS switch has a well established technique, VSANs, which allow virtual domains to be created from within a physical Fibre Channel fabric with each domain having all the capabilities and functions supported by the physical switch. Data storage can be virtualized as logical units of capacity. The SSM, which can be fitted into an MDS switch, facilitates storage capacity virtualization working with partners such as EMC Invista.

Having virtualized the devices and resources, the next challenge is to provision these to applications.

The provisioning and management of a virtualized data center is an important component in Cisco’s Data Center 3.0 strategy.

The last element of a coherent, holistic strategy for the virtualized data center is the automation of provisioning. Having consolidated the data center resources, Cisco can contribute to and facilitate virtualization and is in a position to deliver a resource provisioning solution through ecosystem partner products such as Bladelogic.

The integration of the three platforms creates an area of overlap that has challenged historical organizational boundaries in dealing with the additional complexity throughout the entire lifecycle.

From a site cost perspective, this has translated into improved environmentals—power, cooling, and space—due to the improved hardware utilization and reduced footprint for a static compute demand.

Platform costs have stayed relatively flat. Storage costs have increased due to the need for SAN and/or NAS solutions. Network costs have increased due to the additional network requirements. Software costs have gone up overall from the addition of the virtualization layer required in the stack. Server consolidation, however, has reduced overall quantity and cost.

Organizational costs have gone up. This is because of increased organizational roles, such as virtualization administrators and development roles, and also because of the coordination due to high complexity and high touch created by the overlaps.

By setting up and managing virtual connections, a UCS helps solve the most vexing virtualization performance issues. The Cisco approach to virtualization also expands the utilization rate of each processor in the system. In effect, the UCS extends the concept of virtualization out to create pools of shared memory, storage, and compute engine resources.

Greater efficiencies are recognized as more VMs are added per server. This directly relates to lower power, cooling, and cost per VM.

This chart illustrates how effective a UCS is with regard to cost efficiency. The bar on the left represents legacy rack-mounted end-of-row deployment. Fibre Channel over Ethernet (FCoE) allows for some cost savings by leveraging 10 Gigabit Ethernet networks while preserving the Fibre Channel protocol. However, with a UCS, the combined features of a unified fabric, elimination of unused resources, and ease of management allow for far greater cost efficiency. The end result is a price/performance ratio for supporting virtual machines that is unrivalled in the industry.