Cisco risks de-centralizing data centers to enable expanded UCS platform



By Scott M. Fulton, III 

The idea that processing power should be located close to the client demanding that power is far from new. In fact, it’s now old–dating back to the 1970s when the first “minicomputers” were built for branch offices. As networks became wider and network processing became faster, a great deal of the processing power that had been distributed to branches, returned to centralized data centers–or, in many more cases recently, to cloud data centers worldwide.

Thursday morning, Cisco made an interesting case for that trend to reverse itself, bringing computing power back to branch locations. At a rollout event in New York, the company unveiled a small server form factor called UCS Mini that it says can affordably relocate Unified Computing System applications to the local level.

Borrowing a phrase from Akamai’s playbook, the company called this “edge-scale computing,” which only makes sense if you accept the notion that the place where you do business (formerly known as “the office”) is the “edge” of the network. In the Akamai model, high-capacity multimedia is stationed away from centralized data center locations closer to where customers are demanding them, which is how it came to be called an “edge provider.”

In Cisco’s new model, the old concept of the data center being “upstream” is inverted. “The edge” is wherever computing is taking place, which is where Cisco wants to move its servers.

“How do you address all of the data being generated outside the data center, and how do you basically deliver the computing needs to address that?” rhetorically asked Satinder Sethi, Cisco’s vice president of data center solutions engineering (pictured right). “There are two types of customers with very similar requirements that have similar use cases for this platform. Large-scale IT has standardized their infrastructure in the data center, but there is a need for covering remote office, branch office applications… If you look at small-scale IT, they don’t have the depth of the technical bench to go in, manage this complex assembly of technologies, to deliver their applications. We believe they require something that can power their applications readily, with very simple operations.”

A UCS Mini server is a 6U form factor server, not much taller than your average waste paper bin, capable of housing up to 8 servers with 2 processors each. It ships with a completely new, Cisco-branded management platform called UCS Director that Sethi said promises the ability for admins to comprehensively manage both virtual and physical servers simultaneously, including their running applications, and to pool available storage and peripheral resources among them.

Or, to borrow a word from the previous century, a “minicomputer.”

How would this relocation of servers away from centralized locations benefit a networking company? If banks, financial institutions, and local companies with statewide or city-wide offices were to bring compute power closer to the user (that’s right, folks, away from the cloud), those businesses may be rewarded with lower maintenance and cooling costs. Sethi noted one large customer use case where branch upgrades cost as much as $2,000 per location–making each software upgrade cost millions. Such a cost could be entirely eliminated, he said, if the upgrade process were both distributed and automated.

That gives customers the incentive to invest. And because the operation is distributed, it does give mid-level and some smaller businesses a reason to effectively purchase Cisco networking technology. Whereas in previous years, UCS was billed as a way for networking needs to pave the way for servers, now this model too is being reversed–as a way for servers to pave the way for Cisco networks, which are, after all, its core product.

Except let’s not call them “servers”, Cisco suggests. In fact, let’s get really grammatical about this.

“Our worldview is not necessarily that we are in the server business,” proclaimed Cisco computing systems VP/GM Paul Perez. “We are in the computing business. And there’s a big difference between the server noun and the computing verb. The computing business is a business in which we optimize application environments for performance and for total cost of ownership. As applications continue to evolve, so must our value proposition.”

Related Articles:
John Chambers: Cisco’s move to SDN will take many years
Cisco tries new collaboration, compliance tools for diverse work environments
Cisco CEO Chambers: Most corporations will die unless networks can adapt

Read more about: Satinder Sethi

Jarrett Neil Ridlinghafer
Founder & CEO/CTO
Synapse Synergy Group, Inc.