I've been hearing much recently about private clouds. But is this anything more than a marketing term slapped onto what we used to call a datacentre? Let's go back to first principles.
The idea of a cloud -- which, let's not forget, is just a metaphor that emerged from those Powerpoint diagrams showing the Internet as a cloud -- is that you neither know nor care where your stuff is. It just is, it works and you get it as a service. You can turn it up or down, it's like electricity: a utility. That's the theory.
In practice of course, you may well care where your data lives, not just for regulatory reasons -- it's illegal to send your HR data overseas, for example -- but also because if it's halfway across the globe, it takes time to get here from there, and back again. That's latency.
For example, if you've built a complex application that pulls data from multiple sites or databases, with much inter-communication between them, high latency adds up to a lot of wasted time and Internet traffic, and frustrated users drumming their fingers as they wait for an update. You cannae break the laws of physics, Jim.
But there are ways around this: Amazon's EC2 service for example offers geographical buckets that allow you to keep data close to hand, and any decent service agreement will include some form of detail about where the data lives.
That's a public cloud. What about a private cloud? This is a facility that you own, so you know where the data is, and you can control it. It lives in one of your datacentres -- so what's so cloudy about that?
The key, I hear, is that it's delivered as a service. To deliver the ability to turn the service up and down implies an infrastructure based on virtualisation and a high degree of orchestration -- that's the glorious end-point where virtual servers shuttle around the datacentre to get the resources they need.
And then there's a chargeback mechanism: business units wanting a service will need to pay the private cloud provider -- ie the IT department -- to deliver it.
My problem with this is simple: it immediately puts IT into competition with Amazon's EC2, with Google and with other huge service providers. What are the odds of the IT department being able to use economies of scale to beat such entities on price?
Ask an advocate of this model, and you're told that IT will have to be more efficient. That's much easier to say than to deliver, I would argue. In practice, the BUs are likely rush off to Amazon, credit card in hand, and use them instead.
So can IT become a service provider? Is it set up to deliver in the way that big public corporations do? Does it have those resources? Is this the argument that's going on right now between the CIO and the rest of the board?