Vendor-agnostic data centres: Fact or fantasy?
In some ways, everything about IT is about creating uniformity. Standard configurations. Synchronized patches. 1000 identical laptops shipped to all employees.
We place a high premium on "everything being exactly the same." And for good reason - if everything is the same, management and administration becomes far far easier.
And while we enjoy the fantasy of racks and racks of identically-configured servers humming away in the icy cold data centre of the future, anyone who knows what a data centre smells like (mmm, silicon) also knows that we can, at best, approximate that fantasy - but never truly taste it.
Why? Life is a big, hairy exception.
Exceptions for different workloads. Exceptions for supported software packages. Legacy systems where updates are either impossible or cost-prohibitive. New environments needed to support fun new applications and capabilities. Servers bought by enterprising LOB folks and stored under their desk until they were confiscated like gum during history class. Acquired companies with their own standards, infiltrating your well run data centre. The list goes on and on and on.
In your fantasy data centre, of course you don't need heterogeneity. You're standardised. You're happy. A nice monolithic data centre. But, as noted, you don't get your fantasy. You get your reality - and, as luck would have it, you're in very very good company.
So, onto the topic that interests me - management. Do you need a platform-agnostic management system?
Well, while your fantasy data centre doesn't, the real, live, messy humming one in the back of the building absolutely does. Not just because today, let's admit it, things are something short of glorious. But, also because tomorrow...
Tomorrow, that preferred vendor might fiddle with their prices. Tomorrow, the web team will have a whizzbang new app that requires a different web server. Tomorrow, you'll acquire CheezNCrackers.com to enhance your snack food business, and with it will come another 500 servers of dubious pedigree. Tomorrow, you'll still have to manage all this mess.
The good news is.. today is tomorrow. And the first step to triumphing in tomorrow is accepting reality... and putting that fantasy aside.
- » What automation can learn from DevOps – and why the future is automation with iteration
- » Unravel Data secures $35 million series C funding to help combine APM with big data muscle
- » How augmented analytics is turning big data into smart data
- » Enterprises rethinking their Oracle relationships, argues Rimini Street
- » AWS announces availability of Amazon Managed Blockchain service