Some New York Rangers fans and Washington Capitals fans recently relearned an old lesson: Don’t leave a game before it is over. Someone might score a tying goal with just a few seconds left. Television networks have spent decades honing corny lines meant to keep viewers watching the screen right to the end of every episode.
Data center conference organizers don’t do that, and professionals like me miss an awful lot of good sessions on the last day of an event.
Not this time. The last two sessions of this week’s Uptime Institute Symposium were the conference highlights, both of which I attended. The penultimate session featured Uptime founder Ken Brill and Stanford University’s Jonathon Koomey describing the havoc wreaked by comatose servers in data centers. Brill suggested that as many as 30 percent of all servers could be described as doing no useful work, and he presented server utilization rates that supported his hypothesis. He said
Both Koomey and Brill seemed to agree that the 30 percent figure would prove consistent across the entire installed base of 30,000,000 servers, noting that significant energy savings would accrue from merely upping the utilization rates of this equipment. Koomey said, “The problems that limit energy efficiency in data centers are about people, not technology.”
Brill wore a ridiculous foam hat marked up with a summary of the results Uptime’s Server Roundup and also to celebrate the achievement of AOL, which decommissioned and recycled the largest number and percentage of its servers among the entrants. In total, AOL cleared out 9,484 servers – a 26 percent turnover in server assets. The move saved the organization $5.05 million in utility, maintenance and licensing costs and earned it $1.2 million from asset sales and reclamation. Brill reported that AOL believes it could retire even more servers in another phase. In total, more than 20,000 servers were decommissioned and recycled by program entrants.
Ken’s antics extreme may have been extreme, but years after industry first identified this problem what’s left but ridicule and contests as a way to get these servers out of service? Ken is usually not subtle, and I think he found a subversive new way to make a good point.
Either way, the content of this session was excellent.
Uptime Publications Director Matt Stansbury topped them just a few minutes later, as he shared the results of Uptime’s industry survey.(See Slideshow) His 30-minute talk described a virtual census of the industry or at least as close as I have seen to date. Uptime graciously shared Matt’s presentation with me, and I have posted many of the slides as part of the picture show below.
Take a peak, and decide for yourself where the industry is going. Uptime saw trends towards DCIM, cloud services, and modular deployments, even compared to last year. They also identified barriers to further adoption.
For instance, 64 percent of the 1100 respondents still identify security as the number one impediment to going to the cloud, with compliance (27 percent), cost (24 percent), and lack of management expertise (20 percent) in the next tier.
Uptime found that cost (64 percent) was also the number one barrier to more DCIM implementations. Other barriers were distant runners up.
- Systems are not easy to integrate into existing facilities (31 percent)
- Difficulty populating existing databases (26 percent)
- Existing methods are sufficient (25 percent)
Matt’s presentation included 35 slides, which I am told only skims the surface of the full report.
I am grateful that Uptime allowed me to share so fully with our readers. I plan to do more as this survey includes valuable data.
Regular readers know that I believe our industry would benefit from more open-ness and transparency. Industry research tends to be done by consultants and vendors who understandably tend to treat this information as proprietary after spending lots of time and money compiling it. Uptime is no exception, but the access they gave me for readers is a good step towards the openness I think we need.
Needless to say I was delighted when 451 Research CEO Martin McCarthy and Stansberry allowed me to observe Thursday’s Charrette. My trip had gone into triple overtime!
451’s Hank Seader moderated a morning-long session during which an impressive group of 30 panelists worked through the second phase of definitions geared to developing a process for evaluating IT infrastructure investments. The goal is to develop a common language for executives to communicate across multiple business units and disciplines and to provide a framework or methodology for making effective digital infrastructure investments.
451 Research invited a select group of data center professionals known to the them through their membership, support, or participation in 451 Research programs, and these men and women debated the meaning of finance, opportunity, and risk, which were identified as the hardest of six areas a previous panel had defined in a closed event in Orlando in January, with the others being risk, compliance, and sustainable.
I am not at liberty to say much more about this FORCSS meeting, but I plan to press my opportunity to add at least a little transparency to this process by providing progress reports. I am convinced that Uptime is onto something here, and I want to make sure that the industry understands the final product and accepts or rejects it based on its merits.