Monday 9 May 2016

A Golden Rule for IT Infrastructure

Maybe I'm too old?.. Maybe I'm not aware of some other rule... But when I were a lad, and first started out in a professional IT environment, the key was to keep things available and running, and always the same.  Now, the latter item is somewhat of a grey area today, what do we mean by "always the same"?

Well, I'm not talking about the same box, doing the same job, and nothing changing on it... That is very old school and something you have to think about in the sense of "big iron", when you had only one (for it's day) powerful machine, effectively a main-frame, mini or even just a large server.  So, armed with only the one machine, you were forced to think about keeping the status quo, to keep everything the same, everything happy and you paid lots to have people like me, and hugely expensive support contracts to keep that status quo.

That though is old thinking, that is not how modern infrastructure should work, or even be thought of, now today with powerful servers at lower price points, you can think about "high availability".  This is a concept many many older IT bods and even folks in executive level posts are struggling to swallow.

It means they pay out for many smaller boxes, and they virtualize to spread the work between those boxes.

If one box goes down, the virtual machines, or services being run on it, simply up and migrate to another box in the group.  This is the core concept to modern cluster computing and really only depends on the number of boxes you've bought, or hired, and the back-up/snapshot ability of your storage solution.

So we're talking about boxes, running you a SAN solution, we're talking dedicated high bandwidth between the storage and the processing servers, and then many servers booting locally perhaps, to then load virtual machines over the wire/fiber.

What has all this got to do with the status quo?... Well, very simply in this modern infrastructure, no matter where your virtual machines are running, no matter what their services are, if you've not told your customers in the work-place that a change is coming, NOTHING should change from their point of view.

To change anything at this juncture would result in their feeling, and even spending precious time, checking why something their end is no longer working.. When really it's not their end, something changed in your infrastructure.

So don't do it!

Publish changes, push out a change log, or an update mail or even just have the courtesy to tell them.

And worse still, if a customer comes knocking on your IT office door and tells you they can no longer do today, what they were doing yesterday, don't fob them off with phrases like "if must be your end" or "nothing has changed chap".  If nothing changed they'd not be talking to you, customers don't just get off their seats and show themselves as being in the dark unless something is bothering them.

Your role in IT is not just to provide the infrastructure, but to understand your customers, not all of them are moaning, know nothing, morons... Some are, don't get me wrong, some are, and you are paid to mediate with them.  But then things are changing customers notice, and their frustration will ultimately be taken out on you.

No comments:

Post a Comment