Published: 03 Oct 2013
Learning the names for different metrics isn't enough to improving the quality of your computing and the quality of your data center investments.
In 1958, Oliver Smoot Jr. and his fraternity brothers at MIT measured Harvard Bridge using his body as a measuring stick. The resulting smoot unit is immortalized with markings and a plaque on the bridge, but it’s found almost nowhere else. But people in Cambridge like smoots, which reflect their history and their character.
In fact, you might argue that all measurement systems are arbitrary. Inches, acres, miles, meters, gallons and gigajoules -- their definitions were all chosen. We called the distance light travels in a vacuum in 1/299,792,458th of a second p meter and made it a foundation of the International System of Units (a.k.a. the metric system). But we could have picked 1/300,000,000 or 1/299,876,321 or anything, really.
IT is going through tremendous rebuilding and re-architecting. We’re between a half-dozen and 30 years into cloud, virtualization, bladed and converging infrastructure, Internet everywhere, wireless, mobile, power- and thermal-limited data centers, not to mention analytics and big data. We’ll be actively building out these “new” styles for decades to come -- yet we frequently judge data center investments in these new directions with backwards-leaning evaluations.
TCO, ROI, and such are fine, but they measure the status quo in a world of great and constant change. They also presume that everyone wants basically the same thing: lower cost. In fact, business and IT leaders want a lot of things -- only some of which are cost-related. They want capabilities that make them effective, not just cost-effective. Despite important commonalities, what one shop values isn’t necessarily what another wants -- even in the same industry or location.
We need to take a broader, more systematic view and be more honest about what we really need and value, then make IT decisions accordingly. Consider the following criteria:
Portfolio Fit - We’re building composite portfolios, not individual IT applications and services. How easily and well new tools integrate with and leverage existing tools is crucial. The “best tool” is not the best tool for the job if it isn’t a strong fit with the rest of your toolbox.
Capability Growth - Will an investment help grow your capabilities or fitness? Sure, you want portfolio fit -- but if that’s all you consider, your choices condense and stagnate. “How does this help us branch out and grow?” is the yang to the “How does this fit in to what we’re already doing?” yin. You need both.
Ability to futureproof - We’re always entangled in past choices. We spend or are constrained by choices that, given a second chance, we might have made differently. Breaking that cycle means considering how much a given investment -- or pattern of investments -- locks you in or, conversely, how much it promotes future flexibility.
Trialability - Enterprise software is famously hard to get working, and we glibly excuse failure with “Well, it’s scalable; it has to be hard” or similar nonsense. Consumer companies put a much higher premium on providing ease of exploration and first use. Increasingly, enterprise IT is also asking whether something can be used on a small/trial basis, without massive upfront cost or difficulty. Thanks, cloud!
None of these axes or others such as community, invariability and time-to-value that we don’t have space for here are really new. They’re often lurking in investment discussions, albeit without equitable standing, and are dismissed as being Too Soft or Not Quantifiable. Yet such attributes can be rated. Moreover, they frame an IT shop’s overall ability to execute in ways that metrics such as NPV (net present value) and payback-duration never will.
Modern infrastructure embraces flexibility and systematic coordination. We have a much larger scale and broader vistas than the IT of even five or 10 years ago. We should apply this same flexibility to our data center investment evaluations, judging structural attributes as full peers of financial metrics. To boil this down to a single rallying cry: Smoots!