Q&A

Data center 3.0 is not a product but a design philosophy

Sumit Mukhija, National Sales Manager-Datacenter, Cisco India and SAARC talked about unified computing and virtualization with Rajendra Chaudhary

Unified computing is over a year old and you are still talking about virtualization and data center consolidation. So what’s new in that?

As a technique, virtualization offers a chance to create an infrastructure whereby it is possible for businesses to dynamically provision resources and make them move within the data center almost at will. Although it might still sound a little new to many, we as a company, have been busy solving the data center virtualization puzzle for the last several years. As organizations become increasingly serious about deploying it in their data center, they are faced with some of its inherent complexities in terms of management, security, policies, processes etc. They realize that they are facing the same set of challenges that they faced with physical machines on a much larger scale. In today’s data center infrastructure, resources are separate islands and each island has a different perspective on virtualization. Cisco’s unique value is at the intersection—only the network can bring transparency and visibility across the data center islands to help automation and dynamically provision across the data center infrastructure.

Cisco Data Center Ethernet delivers the architecture for a unified fabric. Can you explain how?

Data center 3.0 is not a product, but a design philosophy that allows our customers to create data centers that have pools of shared storage, servers, networks and application resources, all connected together using an intelligent unified fabric and ultimately gives them an infrastructure to which capacity can be added or removed as and when required.

More specifically in terms of offerings, this new architecture includes new hardware (blade servers), an interconnection fabric, a chassis for the blade servers, fabric extenders and network adapters. It also includes coordinated support and software integration from several of our partners, including Intel, Microsoft, EMC, and virtualization partner VMware among several others.

Can you tell us a bit more about this intelligent unified fabric?

Unified fabric aims at making a server centric architecture into a service centric architecture.

Essentially what it does is that it allows any workload whether it is storage, application or any other kind of workload that you might find in typical data center to be put on any server at any point in time. So here you do not worry about how many Ethernet cards you have in a particular server, the PCI slots and the HBAs (Host Bus Adapters) for fiber channel SAN connectivity. All you need is just two converged network adapters that go and talk to a unified fabric ready switch. The whole idea behind unified fabric is physical consolidation. With this you no longer need separate interface cards for Ethernet connectivity, SAN connectivity or high performance computing. You simply need a couple of converged network adapters and a couple of cables for high availability coming out from every server and they connect to a unified fabric or a FCoE-ready switch at the access layer and from there you have the freedom to connect to any data center resource.

At the access layer, unified fabric results in reducing all the components at least by half if not by more. So you will need half the number of cards, half the number of cables and you have transport flexibility because the same converged network will now be able to take any kind of stress and it requires half the number of switches at the access layer. That’s the bare minimum benefit that it brings to the table. In certain cases, we reduced cabling and other network components by almost 70-80%. Unified fabric is a ratified standard that is making a big impact at the access layer where the servers are connected in the data center.

— Rajendra Chaudhary