Cisco Systems has been making a large deal about converged data center hardware infrastructures for more than two years with its Unified Computing System, as increased functionality gets wedged into smaller components and servers and routers-and as Cisco got into the server business.
Now the world’s biggest IT networking company and newbie data center systems maker is using the convergence headline to explain its 2011 data center software strategy, which becomes physical in the company’s Data Center Business Advantage portfolio.
Cisco on March 30 launched a flotilla of new hardware products, including an application/networking server, several storage switches, something called a “fabric extender,” management tools and a data center management appliance-data center items that refresh a good deal of Cisco’s product lineup.
That’s not all. Cisco also updated its data center operating system, Cisco NX-OS, to run all these new machines.
In fact, the list of new items is so long and detailed that it would be fallacious to try to describe them all here. Thus, we’ll do something we don’t often do here at eWEEK: refer you to the Cisco product announcement for all the details.
The most important new items are the new server, switches and the operating system.
The new Unified Computing System C260 M2 Rack-Mount Server crams even more computing, networking, storage access and virtualization resources into a single box for the rack, Data Center Product Marketing Manager Omar Sultan told eWEEK. This is the box that will run the Cisco OS and contain all the network and storage management tools, among other things.
Cisco also announced that it is now offering end-to-end “fabric-type” connectivity from server to storage with new director-class, “multihop” Fibre Channel over Ethernet (FCoE) in its larger-scale Nexus 7000 and MDS 9500 storage-area network switches. This is already available in its Nexus 5000 switch.
What Is ‘Multihop’?
In the Fibre Channel world, a “hop” is what happens when data moves from switch to switch and the domain IDs change (each switch has its own IP domain). Thus, the new Cisco switches can handle many “hops” as needed to get the data from where it is to where it is supposed to go within the fabric.
The idea of a data center “fabric” can be viewed in a few ways. Most IT people see this as similar to the physical image the word presents: a pliable yet connected structure that can sort of wrap around pools of enterprise data to protect them as they go through the processes they need to endure.
It also can mean that parts of a data center are able to spin up and slow down as workload fluctuates to maintain the best and most economical use of energy for power and cooling.
NX-OS Features Enterprise-Level High Availability
Sultan said Cisco’s NX-OS data center operating system not only runs the unified data center fabric and all its virtual machines, but it also features enterprise-level high availability. Cisco NX-OS has added support for multihop Fibre Channel over Ethernet, which enables multiple points of management to increase the scalability of unified fabric data centers.
Developing, tuning and deploying home-developed operating systems for new-generation data centers and mobile device networks may be a minitrend of sorts.
Hewlett-Packard, a fierce Cisco data center systems competitor, has said in no uncertain terms that it will be using its webOS operating system in all its upcoming PCs, servers, printers and mobile devices. A data center webOS is the next logical step.
VMware has had a de facto data center OS with its ESX hypervisor for a few years, and its parent company, EMC, is folding that in to run its newest storageware. Thus, EMC also has its own in-house data center OS.
Can It Run Nonrelated Systems?
But can this new systems management software run everything it needs to run? Data centers are famous for being nonhomogeneous.
“First and foremost, this is all open and standards-based, allowing customers the freedom to deploy the technology they want to use,” Sultan told eWEEK. “We certainly have a big footprint in the data center, but we don’t do everything. We think it’s important that customers use the vendors [meaning storage, mostly, which is one thing Cisco does not do] that they already have established relationships with.”
This statement stands in direct opposition to what others in the data center market believe: that Cisco’s UCS is really a closed-loop system that is run like a dictator by the company’s network-centric servers.
Since the launch of the UCS back in March 2009, Cisco’s competitors, naturally, have had some interesting reactions to it.
“Would you let a plumber build your house?” Jim Ganthier, HP’s vice president of Infrastructure Software and Blades, asked eWEEK at the time. “Cisco’s network-centric view of the data center is great for bandwidth management, but leaves a lot to chance in terms of service-level delivery as well as data reliability and accessibility.
“The architecture does not unify management, but uses proprietary network-based management structure as the point of control. This is not ‘unification’; this is a change of control,” Ganthier said.