The last great American data center will be closed by the end of January 2020. How do I know this? Because except for a very few companies, the cost of maintaining a data center is rapidly becoming more than what it would cost to outsource.
Yet, it’s not really about the money. This is how everything works.
CIO Wayne Sadin has an interesting way to describe this phenomenon. Early in his career, we worked in a plant that was powered by water and where leather belts were used to transfer and control the power.
“People used to be able to have exactly the kind of power they wanted for their manufacturing,” he told me. “By the time I left, we were using electric motors and had gotten rid of the belts.”
Why? Because even if it wasn’t “perfect” power, what the local electric utility could deliver was both good enough and cheap. But, better than merely good enough, an infrastructure had been created that made commercial power flexible enough to meet almost any need.
New cloud standards
This acceptance, standardization and commoditization of cloud technology are happening now. A combination of napkin calculation, listening to industry chatter and intuition brings me to January 2020 as the conclusion of this phase. You can and probably would select a different number. That’s fine, so long as you actually pick one. This is going to happen.
The data center as the seat of corporate IT is over and cloud computing is its assassin. Virtualization sets up the shot and an IT manager pulls the trigger.
Once you’re virtualizing, it is only time until the light bulb of “we can virtualize anywhere” goes off over someone’s head. That’s when plans start being made for what happens to the space where the data center now resides. And that’s the beginning of the end of the data center as we know it.
Already, there is very little reason for many businesses to maintain their own data centers, but old habits and older mainframes die-hard.
Some CIOs feel the data center must live until either it is fully amortized or the CIO retires, whichever comes last. Other organizations have special needs – like running a mainframe app that would be difficult or impossible to replace – but they are the minority. Still, it keeps my January 2020 prediction from being an absolute.
I am not so stupid as to select the day when the last mainframe will be turned off. Mainframes remain a $5 billion annual business even today. Yet, the generation of programmers who speak Cobol and assembly language is winding down, making mainframes– and other platforms running decades old apps– ever more difficult to maintain.
What replaces the data center?
Physically, the data center will become a collection of switches, gateways, wireless LAN and some sort of backup for the Internet connection the company relies upon to reach the cloud.
I don’t think the desktop is as dead as the rest of the world seems to, so I expect lots of desktop clients to remain for some time. Many of them will continue running Windows XP.
My prediction only works if the cloud lives up to its considerable hype/promise. That seems likely, but if cloud comes up short, data centers won’t disappear as quickly as I expect.
What IT staff remains will be different than what exists today. The technicians will be gone to the cloud providers, programmers will be outsourced and companies will hire people with considerable business and IT skill to make the cloud work as a competitive advantage.
Their job will be almost instant iterations of the cloud apps that run the business, trying to stay ahead of near constant competitive change and fast-moving customers.
That’s how I see the cloud and data centers changing between now and the end of the decade. I’d be interested in hearing what you think.