5 Things I Learned At Gartner Data Center
I spent the first part of this week in Las Vegas at the Gartner Data Center show. I live to tell the tale. Here it is:
1. I still hate Vegas and always will. The smoke. The expensive sleaze. The blasé, graceless service. The aging, overburdened airport. The fact that you tend to spend your entire stay without ever seeing the sun or breathing fresh air. And I’ve come back fluffier (with more avoirdupois, for non-American readers).
2. Cloud, cloud, cloud, cloud. We even seem to be past the previously obligatory “nebulous” puns (finally!), because the conversation is no longer “what is cloud computing” or even “why cloud computing” but “what’s the best way to get there”.
3. In a sign of increasing IT maturity across the board, the “how we get there” talk is pretty routinely multidisciplinary. It has to be. In a traditional world of basically static workloads chained to physical equipment, management and security could be treated as bolted-on afterthoughts with relative ease. This was true even as virtualization was first taking hold. But now that many IT organizations have enough experience with virtualization to be able to envision new possibilities based on a more “logical” type of IT, it’s much more common for people to immediately ask the follow-on questions about monitoring and management, about virtual security and compliance assurance for mobile workloads, and about IT business questions regarding service pricing, metering, billback, and capacity planning—whether internally, with a service provider, or via hybrid clouds. (Gartner analyst Tom Bittman recently tweeted, “One surprise #gartnerdc, huge interest in using hybrid cloud computing (35% in a few years) #cloud”)
4. It was a sophisticated, very savvy crowd. Most knew at least that Cisco has a server platform called UCS; many were from organizations evaluating the platform. Some were more familiar with the architecture than others, and probably the most common question was, “What makes it better?” There are many ways to answer this question, depending on where the person is coming from. There’s the business advantage, succinctly explained in this video:
And then, of course, there are the architectural advantages. I was trying out new ways of talking about them with this audience. Here are some of the things that seemed to resonate:
- Consolidated, streamlined networking. One can talk about unified fabric and simplified cabling, but also about the offload of performance-impacting internal switching on blades and hypervisors with stateless compute nodes and FEX-Link and VN-Link technology.
- Price/density/performance optimization: our extended memory, virtual interface card and portfolio of 2- and 4-socket make it easy to mix & match system components to meet budget and space limitations as well as SLAs.
- Management simplification and abstraction. It turned out to be relatively easy with this audience to draw a straight line from stateless servers to management by calls to dynamically assembled software objects. I did spend quite a bit of time helping people parse where UCS Manager ends and higher-level systems management tools begin. (For the record: UCSM manages everything, physical and logical, about the UCS system—across multiple chassis—but only our system, as it’s essentially a device manager. Provisioning of OS and higher layers of the stack and heterogeneous systems management are handled by our partners.) At which point, this slide consistently came in handy:
5. There’s no place like home. There’s no place like home. There’s no place like home.