I am barely old enough to remember when US “Service Stations” really were – when someone else pumped your gas, and while it was pumping they washed your windshield and checked your oil. Not old enough to have actually used one mind you, but old enough to remember them. Whenever my Mother had to get her own gas – which was not very often – we’d only stop at full service stations, which of course changed over the years, and now that she’s widowed and much older, she’s pumping her own gas because there are precious few service stations left in the world, and none near her house in Cincinnati.
The idea is that the gas is cheaper because they don’t have to pay people to wait upon you. Since gas is now 20 times what it was when there were service stations with attendants, the fallacy of that model is pretty obvious, but gas stations aren’t going to throw up signs saying they’re hiring attendants, that’s for sure. Because it is an overhead no longer required of them, we’ve all adjusted to taking care of our own gas and oil needs.
And that’s where at least some of your IT department is headed, though the economics are different, the model is largely the same.
There are many who dream of a world where business users push a button and get a new image. In some limited instances it may go that far – indeed, at least one university already has gone that far, allowing students to provision their own servers in a virtualized environment with no oversight or interaction on IT’s part.
But my vision is more limited. I don’t believe the vast majority of business departments are going to want the ability to push-button create services because they don’t want to have to retain the knowledge for sizing and security, and they don’t want to have to retain responsibility for the health of the system. I do believe that it is in IT’s best interests to create such an interface though. The ability to instantaneously spin up an instance with the desired amount of storage, the required applications, and the necessary network connections is a huge step toward what I like to call automating the tedious. Let’s face it, a ton of IT staff’s time is still spent doing horribly repetitive tasks that require a base set of knowledge and not much more.
Provisioning a VM if you are using the “copy VM” function in VMWare is not rocket science. The difficult part is getting the original image set up correctly and then changing the IP information for the copy to be unique. The rest is nearly push-button. So automating the entire process to take the original and do everything including assigning unique IPs would save you a ton of time. Automatically adding the image to a Load Balancing Pool like those provided by an F5BIG-IPLTM Application Delivery Controller is even better. Of course when you “push button” you’d have to have some questions – like how much storage is needed, if it should be added to a load balancing pool, and if so which one, what the subnet should be, etc. But it can be done via scripts, our iControl will let you create the node (given it's IP address), add it to the pool (given a pool name), and even create the pool if it doesn’t exist, should you want such functionality.
That frees up valuable IT resources to focus on the business problems, not the technological barriers to getting them implemented. And moving forward that is where you want your focus to be. Security, access control, business problem. While all of us love to be the IT geeks, the business needs IT to grow beyond that, and has actually be pushing for such growth for a long time. And the technology has reached a point where you can. The above referenced script can even be extended – in some environments – to include spinning up VMs in the cloud if you need to shift some load out of the datacenter. Again assuming you want to.
So set your uber-geeks to setting up an automated environment. Challenge them to make it as smooth, automated, and actually useful for your specific environment as possible. Include developers and systems/network admins, because this problem requires a bit of both, and reward them well for achieving an automated solution.
The result? You won’t have service station attendants any more, you’ll have personal trainers that can focus on the needs of business units and build them solutions. In theory any IT person can replace any other IT person (this is of course bogus, but ask your business users, they believe this – or at least most of them do), but no one can replace the guy that keeps their specific application running.
And if you did it right, when you bring new people on, you won’t have so much “oh yeah, and that system uses a different naming scheme because of a dependency on a purchased package…” type secret sauce for them to learn, because it will be accounted for in the automated system.
Raise your game a level. Stop being IT and start being our IT – the people the business relies upon to solve their problems.
Of course you won’t get there overnight, and of course you won’t have a completely automated system with no oddities in it, our datacenters have been growing for decades, and there’s always some weird piece of software or hardware to maintain, but if you can get to 90%, that’s a lot of manhours preserved for other uses.