There's a world of difference between what the average IT person does today and what they'd have done ten or fifteen years ago. So where's it going? What will we need to teach the next generation of data centre staff?
The shifting focus
The big difference between yesteryear and now is the growth in managed hosted services. “The cloud” is the name most recently attached to this concept, but the idea has been around for years; that said, it's far more popular than it used to be and its popularity continues to grow.
This means that the focus of the support effort within the average organisation is much more on the application. Instead of buying servers we are running up virtual instances in the cloud, and this means we now have to worry far less about how server hardware works and far more about how the applications perform, and how to troubleshoot them and optimise their performance.
We still need to understand the operating systems and how to design the setup such that you can run OS and application updates without unnecessary downtime. It's only really the server hardware that's losing its focus - along perhaps with the licensing aspects of the operating systems as the service provider handles that.
There's lots of stuff staying the same
It's very unusual for everything to move out of the data centre and for the DC to close down completely, and so you'll still need to have some server equipment and the appropriate network infrastructure.
You will also need to be able to use and understand the monitoring tools that come with the platform, and you still need to understand the apps and keep track of their licences unless you've gone on a sizeable SaaS mission and decided to let the provider do that bit too.
It’s the same with security, firewall rule design/management/audit and the processes associated with user adds, deletions and changes. All these staple knowledge requirements aren't going away any time soon, and so they need to be taught to the next generation.
Three things you must learn
So we've said that there's a bit of a move away from server hardware and a lot of stuff that already exists. Not a big deal, really. But you need to ensure the next generation understands these three big areas.
I come from a generation where you had to design and use applications efficiently, because the resource they had available was incredibly limited. You simply couldn't get away with writing inefficient code because it would just overflow the minuscule amount of available processor and memory and crash horribly.
These days you can get away with murder in writing horrible code, and the super-fast machine it's sitting on will save your sorry ass (as will the framework within which the code's running by cleaning up your garbage even though you made a sorry mess of memory allocation and de-allocation).
If you wanted to buy a server, you had to justify that to someone. As well as the cost of the box there was the OS, the maintenance contract, the data centre space to put it in, the power provision, the whole nine yards.
Today if you want a new server you click “New instance” (or perhaps “Clone instance”) in the management GUI of the cloud setup. It's super-easy, which means the temptation to do stuff inefficiently and just throw processing power at the problem is greater than ever.
The initial cost of a virtual server in a hosted environment is the square root of bugger all, and so it's easy to run up loads and leave them humming along. But every minute a virtual machine is running is costing you money. Five VMs cost five times as much. 50 cost 50 times as much.
VMs in a cloud setup are very cheap - until you add up the cost of what you have, when it's common to get a bit of a surprise. A few pence per hour doesn't sound much until you've got 50 more virtual servers than you budgeted for.
To be fair to the cloud providers their financial reporting is considerably better today than it was even a couple of years ago and it's relatively straightforward to track costs and project spend for the year. But you still need to be judicious and monitor your costs closely, or you'll have a nasty shock before too long.
Customer management When your servers were all in the data centre, you could drop everything and rush down there if something important went pear-shaped.
Although most companies have some kind of internal service level agreement between the IT department and its customers (the users), most also have the ability – perhaps on an informal basis – to escalate a super-urgent problem (if the finance system turns up its toes at year-end, for example).