All tech becomes obsolete at some point.
I've had the pleasure of discussing IT with CIOs across a number of industries for many years. Surprisingly, only a few of them have told me that they worry about exit strategies for their technology. As simple as it might sound, at some point, all technology becomes obsolete. CIOs generally take into account the lifecycle of the gear they buy, but rarely do they consider a strategy for getting out of the technology they're buying into.

Consider this. Mainframes were once the bomb. IT spent a gazillion dollars (technical term) to build their business around this technology. At some point, which we won't argue, mainframes became … ahh … not the thing any more. So, we've spent the last decade or three re-inventing the wheel. We did this partly because of the gee-whizz factor of newer technologies, but also because they have compelling business and technical factors. The mini-computers of the 70s, the open systems of the 80s and 90s, and the Internets of today all provided compelling reasons to migrate. But, rarely were plans put in place to build agility into IT infrastructures to accommodate for change. I personally know of no IT organization that had built their mainframe, mini-computer, or even big UNIX boxes around the idea that at some point they would become obsolete. Instead, processes, both mental and operational, were built around these technologies being around forever. But "forever" is actually not infinite in IT terms – it's usually 5-15 years.

Let me give you an example. Back in the middle ages of UNIX, I was a systems administrator. One bright company decided to re-invent a mainframe concept and developed a pretty cool hierarchical storage management (HSM) system. It used huge hard drives (300MB) and a robot with optical MO media. A fully configured system gave you 600MB of HD space and a mind-numbing 20 GB of total storage. To me, it seemed like infinite storage. We turned this system into the NFS server for all of our home directories. We built cool processes around making sure that users didn't do boneheaded things, and generally made sure that ALL of our stuff was in that box. Well, 600MB became nothing, and 20GB became even less shortly thereafter. But, all of our processes were tied to that box and how it managed space. However, because we hadn't planned for it, we couldn't move off this system. Doing so would have taken more (down)time than we could afford. We kept that box running until we literally could not. So, with great pain, we had to tell our users that they would not be able to work for several days while we got them to a new, shiny, faster thing. It cost us much more money to do the move than it did to buy the next three generations of technology. Oops.

Turns out that my experience with this HSM system is still happening today, some 20 years later. Really. We constantly run into customers that have bought into some storage vendor's vision of archiving, or some server vendor's vision of operating systems, or whatever – only to find out that they are now locked into that vendor's vision and technology. Right or wrong.

There are many answers to this problem, but one of the simplest, most overhyped, but underutilized is open standards. The concept is rather straightforward. If you use something that is not vendor locked (not tied to a specific vendor's implementation), then it's highly likely that the next thing will have a way to get your data/process/etc. out and put it into the new thing. If you're vendor locked, then generally only that vendor can realistically get you out. And guess what, they're going to get you out of that old thing and into their new thing. That is if they come up with a plan to do that and if they're still in business. And even if they do come up with such a thing, you're most likely locked into their next thing. And that extends the cycle.

Plan for obsolescence
So, when you're planning the lifecycle of the gear you're buying, please make sure to account for the lifecycle of the technology that you're buying. It's easier said than done, but with careful study and planning, it's doable. And you're probably going to save yourself a ton of money.