If you've made it this far, we've on an interesting journey together -- exploring concepts around an internal "information utility" for our organizations.
Yes, some pre-reading is required: an initial discussion here, concepts of efficiency here, building control models here, and -- of course -- how we might think about choice.
In this post, we come full circle, and talk about how organizations might get from here to there.
It's All About The Journey
During 2009, I probably spoke with 200+ IT organizations around private cloud concepts. Once they understood the concepts, they agreed -- this was a fine destination to chart a course towards.
But how to get there?
If you remember, three things had to change at the same time to gain the benefit.
The environment had to be built differently -- dynamic pools of resources.
The environment had to be operated differently -- low-touch and zero-touch operational models.
And the environment had to be consumed differently -- convenient consumption, pay for what you use.
That's a lot of change to bring about in a relatively short period of time!
What was needed was a place to start.
And -- if you agree with some of the concepts I've shared around "information utility", you'll recognize the parallels.
Building The New Thing
When we got into the heat of a private cloud discussion with an enterprise IT customer, we quickly learned to "target the unmet need" as a starting point.
Simple question: who are your crankiest users with unmet needs?
The answers varied based on industry -- my developers, my engineers, my data analytics team, my power users, etc. etc. It was a rare case where we didn't find some group of users who (a) were important, and (b) weren't getting what they needed from the enterprise IT group.
We simply suggested that meeting their needs with a private cloud might be an interesting place to start. After all, they'd be the most willing to try something new, yes?
Build them a private cloud. Use the new technology. Use the new operational model. Use the new consumption model.
Call it a pilot, a proof-of-concept, a skunk works, an experiment -- anything to set people's expectations appropriately.
Don't use the same process and methodologies that built your legacy: carve off a bunch of bright people and give them the space to do something a bit different than everything else.
And, surprisingly, that's what happened in many cases -- with spectacular results.
As an example, one of the private clouds at EMC runs about ~1200 VMs and about ~2 FTE doing the entire thing end-to-end.
As we all get more confidence with the technology and operational model, expand the use cases to more traditional requirements. But don't compromise on the efficiencies you've created!
The same sort of path might be attractive for building the "information utility".
Where To Start?
Who in your organization seems to have a bottomless appetite for storage?
Is it your developers? Business analytics team? Maybe your VMware farm is eating storage like popcorn?
If you could build an "information utility" for this group, what would a "win" look like?
Would you measure delivered cost? Time to provision? Ability to vary service levels and cost? Provide granular charge back? Pass the compliance audit in 15 minutes or less? Implement self-service data protection and recovery?
"Good" means different things to different people -- what does "good" mean in your organization?
My advice?
Get started.
The technology is there, or will be very soon. Once the technology is there, it's people / process / politics that stand in the way of gaining the benefits.
And --trust me -- the technology will be there long before you fix the other stuff :--)
Comments