Moving to the Private Cloud: Rethinking or Incrementalism

My colleague Javier had an interesting article in his blog about the Greenfield Enterprise where he looked at what one would build if one started building IT entirely from scratch. It’s an interesting article and you should read it if you have not. Keep in mind that Javier's agenda is around SaaS and this article was a setup for answer, as he indicates at the end.

I've been bothered by a smaller but related question, whether moving to the Private Cloud requires somewhat of a rethinking and re-architecture or whether it can be done with some level of tweaking and integration around existing tools and infrastructure. The catalyst for this question came from a customer visit from a few months back. The IT folks at this customer had started a significant pilot project to provide a "cloud model" for their end customers/consumers. My expectation was that they would take a subset of their use cases, some new hardware, create a fully virtualized environment and start experimenting with the use-cases, new tools, and processes to figure out how they would get to a "cloud model" that solves their end-users needs for agility and reduces the admin overhead and opex costs.

I was surprised to hear what they viewed as their constraints

  • They had invested a lot of money and time in setting up the tools and processes from two different ESM vendors to cover their current physical datacenters. They felt strongly that the final solution must be based on the tooling that they already had in place with integration of some new tools.<br /><br />

  • They expected the new "cloud" and tools they were building to encompass all the physical environments in addition to the virtualized environments.

Regardless of the end result, I would have at least started the thinking process with a blank sheet of paper, not cluttered by all my current constraints and sunk cost. Interestingly, on this customer visit we first met with a group of architects who gave us the scenario above. Later we met with their operations group. This latter group seemed to recognize much more clearly that there would need to be a fairly clean new start if they were going to achieve significant change and efficiency.

I'm sure that this is a question that is doing the rounds at many companies. Is this black and white or are there many shades of gray. Wonder where people stand on this one?

0 Kudos
1 Reply

Attrition is a beautiful thing.

I say this because I believe without a doubt it is the easiest and most likely approach to cloud architecture, wether it be public, private or a hybrid cloud model, applications and infrastructure running in our data centers today are not built to take advantage of cloud architectures leaving us with attrition as the most likely method businesses will implement cloud infrastructure.

I could not agree more that companies need to think more about how they will embrace cloud and what it means to their specific business rather than how a given tool chooses to deliver cloud services. A blank sheet of paper or in many cases the infamous white board is exactly where one should start.

The "greenfield" notion is one that many of us dream of, thankfully it is one my organization gave us a chance to take head on this year. We chose the name "Clean Room" rather than "Greenfield" since we essentially carved out existing data center space into a clean room to implement a private cloud model that spans multiple data centers and redesigned infrastructure from the ground up. Many organizations will be faced with similar projects and need to ensure they check any previous infrastructure management methods at the door.

Implementing clouds is one thing, designing and hosting applications that are built for them is something entirely different. I honestly think it will be application integration with clouds that we all struggle with and it is not until we integrate applications that we reap the true benefit of cloud.

Thanks as always for sharing your thoughts Ben!

Charlie Gautreaux


Charlie Gautreaux vExpert
0 Kudos