This article can also be found in the Premium Editorial Download "Modern Infrastructure: The incredible shrinking desktop: Have Apple and consumerization finally broken the back of Wintel?."
Download it now to read this article plus other related content.
When it comes to modernizing IT strategies and infrastructure, the first, worst and most common pitfall is being distracted by early adoption of shiny new gear and ignoring the need to rethink how that new gear is sourced, organized, managed and used.
It's easy to just keep doing what you were doing. But that leaves most of the potential value on the table. Modern infrastructure isn't about having the latest and greatest gear. If it were, everyone would just buy it, and things would be ducky. The real opportunity is improving IT's attributes and outcomes: How fast it provides answers. How fast it moves, changes and adapts. How much it costs. How well it supports and grows business opportunity. Changing those things requires re-architecting the data center -- and indeed, the greater IT supply chain.
We saw this, for example, with the first generations of blade servers. They provided a cool new form factor with exceptional compute density and great plug-and-go flexibility. But many blade adopters didn't focus on making a fundamental shift toward modularity and flexibility. They kept putting local storage on every blade, and some insisted on managing each blade individually. By conforming to the legacy of rack servers, it was almost as though they weren't blades, and certainly not part of a shared infrastructure. Then, a half dozen or so years later, the use of virtualization, shared I/O and shared storage started picking up steam. As a result, blades finally came into their own: they started providing, in broad practice, the modularity and effectiveness they had been capable of years before.
Not everyone had to wait five years, though. Some early adopters got it long before the mainstream because they shifted approaches, rather than simply re-fighting old battles on fresh fields.
Modern IT infrastructure and practices -- cloud, blades, virtualization, scale-out, agile development, converged equipment, everything as a service, DevOps, etc. -- combine to make a very different landscape and set of capabilities than what IT has had before. Using those tools and technologies effectively requires different ways of working. Not everyone is comfortable with those new approaches (or with new approaches in general). Thus we have a wide built-in chasm between most peoples' thinking, practices and expectations and what's needed for a truly flexible, network-savvy operation.
Minding -- and bridging -- the gap
I'm often asked to do gap analysis. After discussing what a client's already doing, plans to do or wonders whether it should be doing, we get to "So, what are we missing?"
"What are we missing?" is the best question because most organizations already do a reasonable-to-good job in most areas. We can tweak -- occasionally, outright fix -- things that could be done better. But the most interesting risks, dysfunctions, weaknesses and opportunities tend to lie in the things they're not doing; frequently, in things they're not even thinking about.
When it comes to modernizing data center strategies and operations, here are some common gaps I've encountered:
Static data. The failure to rethink the data layer -- the combination of storage, databases and transactions -- with the same vigor and depth as organizations rethink servers, insourcing (INSERT LINK to " The benefits of insourcing data center operations"), outsourcing and other issues. Virtualization's buildout over the past decade showed that getting the data layer right is at least as important as getting servers and other infrastructure right. IT is data processing, after all—and even more so in the age of analytics, real-time updates and big data.
Management inertia. Too many shops continue an equipment-centric view of how things are -- and should be -- managed, rather than a service- and application-centric view. A similar gap is continuing to think of development and operations as separate, ne'er-the-twain-shall-meet activities. That "throw it over the wall" mentality eventually stymied progress in manufacturing, and it doesn't work in IT. You can't be efficient if you can't share, and you can't be agile if you don't eagerly cooperate.
Failure to automate. As IT deals with much more flexible and varied underpinnings -- thanks, virtualization and cloud! -- and is called upon to ramp services up and down rapidly, it needs to move away from building everything by hand to building recipes. Shops on the forward edge of scale-out IT are already deeply into deployment automation, because there's no real alternative. It's time for the rest of IT to move in the same direction.
Modern infrastructure isn't a synonym for early adoption. Late-model equipment nicely advances performance, integration and efficiency. But rebalancing how we use IT to take advantage of the leverage that virtualization, cloud and convergence gives us is a critical companion to upgrading the gear. When you're modernizing, don't just focus on "New! Shiny!" Getting the full benefits requires at least equal attention to, "How do we need to change how we're doing things?" and bridging the gap between how you used to work and the new way.
Let us know what you think. Write to us at firstname.lastname@example.org
This was first published in April 2013