05.11.10
Don’t think about servers, think about Workloads.
It used to be so easy – if you were a techie – to describe an environment. You’d simply list out the numbers and types of server, the OS versions running and the server-side apps they’ve been dedicated to. We’d even call them “The SQL server, the Exchange Server, the AD controller…” These physical devices were know by the service they were running.
And if you wanted to add another service you’d, more than likely, add another server or two to run it. This was the easy way, the most trusted and risk free way because we knew that the requirements would only ever increase and we were building in future-proofing. We had to because these things had to last, to run and perform for 3-5 years! This was really about money, IT departments speculated at the start of each financial year as to what would need to be replaced, upgraded and what new systems or services would be required. They got their budget, or a part thereof, and then did what needed to be done, perhaps picking up a few ‘goodies’ along the way – because that’s what we did!
So corporate server rooms and datacenters and even small businesses with the typical too-hot, under-powered cupboards had server hardware spinning away just waiting, ever waiting, for the time when their capacity would be realised. Of course the reality is that, that time never really came. Servers went end-of-life and were retired, re-purposed and ultimately scrapped (or sold on eBay), budgets were re-negotiated to replace them with the newest models because “they will provide significant future-proofing” and the cycle of speculation began again.
The times they are a-changing though and the move is away from the old server-per-service model and we are now starting to talk about workloads. The Database workload, the Exchange workload, the Web workload, the Application workload… All are the new expressions of what needs to be done and none talk about the hardware underneath – and neither should they!
The correct and only way to deliver services is in the most resilient and efficient method possible. This means taking a number of physical resources sufficient to provide service continuity in the event of failure and making them run these workloads. If that sounds kind of wooly and fluffy, well that’s because it is! Server hardware should not, now, be the definition of a service infrastructure it should simply be the mechanisms upon which the services are built.
Infrastructure-as-a-Service (“IaaS”), one of the three main “Cloud” definitions, means that I can define what I want in terms of workload and get it, I don’t need to worry about the numbers, the networking configuration, the will-it/won’t-it nature of any fail-over I’m expecting or speculate upon the “future-proofing” my hardware spend is assuring. This is because I have no hardware spend.
If my requirements are for a database and application workload I can simply buy the horsepower I need to support that DB/Application combination and use it, usually within a few hours. OK I don’t have the fun of specc’ing up a server set and un-boxing the kit when it arrives but they were short-lived pleasure anyway!
For smaller businesses and organisation there is the issue that this is going to be a hosted service, out there in “The Cloud” or simply on-line if you’re me! And this might pose something if an initial barrier to adoption. That’s OK! Really it is, there’s no-one telling you that you have to move your “stuff” out there today, or even tomorrow. IDC has predicted that we’re in the first phase of an adoption cycle which may last up to 30 years! It’s fine to hang back, make the most of your current investment, make sure you understand what it means to adopt an on-line/hybrid service model and make sure your strategies – both technical and business – are in place and ready for when you make the move.
For larger enterprise businesses though, it’s perfectly possible to build an IaaS infrastructure in-house. The technologies are available today and, depending upon the age of your kit, you might not need to spend a huge amount on server hardware to make it happen. The keys to delivery are virtualisation and automation and the tools are ready now to help you deliver this “Private Cloud”. It’s a change in thought processes for any IT Team but it’s also the right way to think about your technology services and how you consume them.
Remember the word is Workload, the method is up to you!
