Original URL: http://www.theregister.co.uk/2011/01/07/dc_introduction/
The optimised datacentre
Introducing our new topic
If you're a datacentre manager then you know just how tough it gets. Whether the customer is a small business down the road or a large enterprise halfway across the globe, whether you work for a service provider or in an in-house datacentre, it all amounts to one thing: an unsquareable circle.
On the one hand, everybody wants everything yesterday, perfectly formed, and all on or under budget. Perhaps that's the relatively easy part. Because despite what the vendors say, nothing goes exactly to plan. Maybe the technology works as it should - even though that's not a given - but when you find the delivery vehicle has broken down halfway up the M1 just when all the rack space has been cleared, and the plan is to have the new servers up and running by tomorrow morning, you wonder what the point of planning is. At this point, you will, clearly, have a plan B....
"What's the answer? More technology? Cloud computing? Resign?"
At the back of your mind is the vexed issue of power. Not how much you personally can accrue but how much you can cram into the datacentre. You've a couple of major projects coming down the line - a new corporate website and a major network upgrade - and you need to know that the IT gear will remain inside the power envelope. But how much can you trust can you place in the specifications?
Meanwhile there's a number of problems you can see coming down the road. For instance, the daily backup takes longer every day. There appears to be no end to it, to the point where forward projections show that, inside a few weeks, the backup window will close entirely. Something has to give, as there's no stretch in the budget for a major storage upgrade.
What's more, IT staff still spend too much of their time firefighting - keeping the lights on - when you'd much rather they expended their energies thinking about and planning to make existing systems work smarter. Instead, systems persist on being slightly incompatible with each other, despite what the vendors' glossy brochures say - or in many cases entirely incompatible: ever tried to cram an HP server blade into an IBM chassis?
Technologies such as virtualisation that promised a brave new world has delivered savings in hardware and maintenance, but there's a whole host of other issues. Have you ended up with a more complex version of the old system? Despite virtualisation's power savings, the box that houses it all, the datacentre, is so full of stuff that it sucks as much power as the power company can deliver, and would take more if the electricity company had some to sell.
Good news? The good news is that you have an experienced team of dedicated hard-working staff who, given the resources they need, can usually make things work in the face of adversity.
More good news? You have at your disposal better and more capable technology than ever before: servers offer huge performance from multi-core processors and vast gobs of memory, storage is cheaper than ever yet sports an ever-growing list of features that could make it cheaper to run, and the network just keeps getting faster.
Yet all this never seems to resolve the fundamental issue of extracting more out what you already have rather than buying new stuff. So what's the answer? More technology? Cloud computing - whatever that means? Resign?
We aim to help you forge a path through the thickets that confront you, and move smoothly towards the sunlit uplands of automated datacentres where all you have to do is say: 'make it so'. Well, you can dream.... ®