Monday, August 2, 2010

Internal clouds: with the right approach, you might be more ready than you think

If you saw the headline “You’re Not Ready for Internal Cloud” making the rounds last week, you might have thought Forrester analyst James Staten was putting the nail in the coffin of the private cloud market. Or maybe you just thought he was channeling Jack Nicholson’s Colonel Jessep from “A Few Good Men” (yelling "You can’t handle the truth!”…about cloud computing?).
It turns out neither is the case as you read that Forrester report and look around at what customers are doing: internal clouds aren’t to be dismissed. And customers can handle the truth about what’s possible here and now.
James deserves congratulations for once again crystallizing some key issues that customers are having around cloud computing – and for having some good advice on ways to deal with them. (It’s not the first time James has been ahead of the curve with practical advice, I might add. For some other examples, check out the links in my interview with him from a few months back.)

It turns out James hides one of the most important messages about “being ready for internal clouds” in the subhead of the report: “…You Can Be: Here’s How.”
Here are a few things that struck me in reading both James’ report and commentary on it last week (Forrester clients in good standing can read James’ write-up here):
The industry expects internal (or private) clouds to either be dead simple or impossibly complex. The truth is in the middle somewhere. If you’re trying to take what you already have and get those IT systems to begin behaving in a cloud-like manner, there’s certainly some work to do. Rodrigo Flores of newScale paraphrased Staten’s reasoning in his blog here as this: “Cloud is a HOW, not a what. It’s a way to operate and offer IT service underpinned by self-service and automation.”

Creating an internal cloud includes both technological changes and alterations to the approach that IT departments are taking to managing and distributing their infrastructure and resources. Because of that, it really isn’t possible to buy an internal cloud out of a single box. Plus, I’ve heard customers that are very interested in building on what they’ve already invested in. That requires an IT department willing to invest as well, in things like time, effort, but most of all, change. But there are some comparatively easier options. More on that in a moment.
You can’t just turn to your vendors for a simple definition of cloud computing that’s going to automatically match what you want. You have to get involved yourself. James has been one of the industry watchers who isn’t afraid to call vendors out when they are using a cloud label on their offerings without good reason. He notes that the term is thrown around “sloppily, with many providers adding the moniker to garner new attention for their products.” He warns that you can’t just “turn to your vendor partners to decode the DNA of cloud computing.” His comments match what I say in my Cloud Computing 101 overview. I believe the definition of cloud computing (and in this case internal or private cloud) is less important than actually figuring out what you can and should be doing with it. And that’s something that IT departments need to work on in conjunction with their vendors and partners. You can’t just take their word for it.

You do need to be pretty self-critical and evaluate your level of readiness for an internal cloud. In his article about James’ report, Larry Dignan from ZDNet notes that “there’s a lot of not-so-sexy grunt work to do before cloud computing is an option. IT services aren’t standardized and you need that [in order] to automate.” The Forrester report has a check list of things you should think about to decide if you’re ready to take some of these steps.
One of the highlights from that list is the importance of sharing IT infrastructure among multiple groups. As Dignan says, “various business units inside a company are rarely on the same infrastructure.” However, writes James, “the economic benefits of cloud simply don’t work without this” kind of sharing.
A while back (over a year ago, in fact), my blog co-contributor Craig Vosburgh posted some of his thoughts about what you must be ready to do from a technical standpoint to really operate a private cloud. It turns many of these are pretty complementary to James’ list. Craig mentioned the willingness to be comfortable with change of a style and scope that IT shops haven’t been used to before, given that workloads will possibly be shifting around from place to place at a moment’s notice, along with a number of other, more technologically-focused considerations. Whatever list you use, the point is that you need to have one. And you need to make sure you’ve thought through where you are now and where you’re going.

One measure that James spends a good portion of his paper talking about is the level of an organization’s virtualization maturity. His belief is that this will have a strong effect on their ability to make an internal cloud a reality.
Is this a useful thing to measure? From what I’ve seen, yes…and no. Virtualization maturity is important, but greenfield opportunities give you the chance to start from scratch with something simple and show big gains. James mentions several ways to actually deliver internal clouds. Only one of those involves progressing through all stages of a virtualization & automation project involving you existing, complex infrastructure for your key applications. He notes (and customers have, too) that you can actually start with something much simpler.
After all, as Rodrigo Flores noted in his blog, “why get stuck solving the hardest, most complex applications” when, in fact, those tend to be pretty stable, don’t really move the needle, and are often under the most intense governance and compliance scrutiny.
This is where a turnkey cloud solution can be a big help (and, yes, it should be noted that CA Technologies’ 3Tera solution falls into that category). Instead of having to move through a slower set of evolutionary steps around the systems and applications that you already have running and don’t want to disrupt, the other option is to start something new. Pick one of those new applications you’ve been asked for that’s not a bet-the-business proposition and try something that gives you a big leap over what already exists or the standard approach. If it fails, you learn something at minimal risk. But if it’s successful, you have a powerful new model to think about and try to replicate in other areas.

And, enough with the excuses. The bottom line is that trying one of these approaches is going to be critical for the IT department’s future. There are certainly very good reasons that internal or private clouds are going to be difficult. But I believe IT is going to be a lot like James Urquhart envisions in his recent post: it's going to be “hybrid IT.” For some things, you will leverage applications and IT infrastructure from inside your firewall; for other things, you'll get help from outside. There will be a need for managing each piece, and to manage the whole thing as a coherent IT supply chain. The more experience that you and your IT shop have with even a piece of this early in the process, the better choices you’ll be able to make from here on out.

In either case, take a look at James Staten’s report if you can. It’s helpful in thinking through these and a bunch of other very relevant considerations around internal clouds. But maybe its most useful role is to serve as a bit of an internal rallying cry, helping you get your act together on internal clouds.

Who knows? With the right approach, you might be more ready for internal clouds than you think.

No comments: