How many more reports pointing out the productivity gap between the promise of IT and the results actually seen in the economy are we going to have to listen to before we get the message?
Whatever we’re doing, we’re not doing it in a way that unlocks performance.
The IT industry has also bifurcated: there’s the exciting part chasing the consumer, and the enterprise part which, by and large, is trying to hold onto as much of the past as it can.
Meanwhile, new possibilities, like cloud computing, get jammed into the same old strait-jackets. “Come use our cloud solution — it runs the same old fairly monolithic application!”
Let’s be clear about what the cloud approach — as opposed to the server-based approach, or its client/server predecessor, or its centralized processor before that — actually means.
It means a fully virtualized environment, in which changes in workload — a burst of transactions, for instance — can be responded to by creating a few more instances that’ll respond to the load.
It’s about responsiveness. When you’re in a room and it gets darker outside — as with this week as Hurricane Sandy burst onto the scene — you turn on a few more lights to make up for the light you’re not getting from the window. Your hydro bill goes up a bit, for as long as you need the extra illumination.
That’s what our systems should be like. That’s why hybrid clouds make sense: clouds in our private environment (whether on-site or in a co-location facility or in a managed service provision facility) coupled to public cloud facilities.
But they only make sense at the end of the day if our application workloads are designed to use that power. You might start with hypervisors in a virtualized infrastructure shifting power between workloads. But where you want to get to is quick start/quick stop instantiations meeting demand dynamically.
Now … how many of your applications are designed to work that way? Or are they designed to simply have more hardware in the same instantiation thrown at them (and so you need to purchase for peaks) when under load?
Major package vendors know they have to head this way. But you can only get there with them if you can get off the release you’re running and get up to the latest versions as they come out.
Stick with the philosophy of “everything needs a package, and every package must be bastardized to fit us” — the old “the user always gets to avoid change” model of the past half century of IT once System/360 made transportable and upgradable workloads possible — and you’ll never get there from here anytime soon.
CEOs keep saying, whenever they’re asked, that IT has to start generating new value for the enterprise. Being a cost efficient player isn’t good enough any more.
If you’ve got consultants and systems integrators reinforcing the past by helping you modify packages, pick packages rather than recognize a value-generating differentiation better served with in-house code, or helping you just ride your workloads into the future as is, you’re missing not only the value-generating opportunity, you’re setting yourself up to be outsourced out of existence.
There isn’t a CEO or CFO in existence who won’t look at people who won’t deliver that won’t find the cheapest way to stabilize a cost they can find.
To make the cloud deliver, literally everything — every package, every practice, every vendor, every architectural assumption you have — is up for review and change.
Run with it, and build a new future for yourself. Run from it? You won’t just be risking your own role.
Those enterprises that leverage the power of the hybrid cloud are going to eat their competitors alive.