Summer vacation is really a bad idea.
At least, that’s what TIME Magazine reported a few weeks back. Despite our glorified, nostalgic memories of endless hours on the tire swing above the old water hole (or, more likely, trying to find towel space on a lounge chair by the gym’s overcrowded pool), apparently kids forget stuff when they aren’t in school.
So, now that everyone’s headed back to the classroom and hitting the books again, they’ve got to jog their memories on how this learning stuff worked.
Luckily, as working adults who think about esoteric IT topics like virtualizing servers and actually planning cloud computing roll-outs, we can say this is never an issue. Right? Anyone? Bueller? Bueller?
However, with VMworld imminent and people returning from vacations, it’s a good time to reiterate what I’ve been hearing from customers and others in the industry about how this journey around virtualization and cloud computing goes.
Some highlights (take notes if you’d like; there might be a short quiz next period):
Rely on the scientific method. You’re going to hear lots of announcements at VMworld next week. (In fact, many folks jumped the gun and lobbed their news into the market this week.) In any case, be a good student and diligently take notes. But then you should probably rely a bit on the scientific method. And question authority. Know what you need or at least what you think you need to accomplish your business goal. Look at any/all of our vendor announcements through that lens. You’ll probably be able to eliminate about two-thirds of what you hear next week from VMware and all its partners (and, of course, I realize that probably includes us at CA Technologies, too). But that last third is worth a closer look. And some serious questions and investigation.
The answers aren’t simply listed in the back of your textbook. Meaning what? Well, here's one thing for starters: just because you’re knee-deep in virtualization doesn’t mean you’re automagically perfectly set up for cloud computing. Virtualization is certainly a key technology that can be really useful in cloud deployments, but as I've noted here before, it’s not sufficient all by itself. The NIST definition of cloud computing (and the one I use, frankly), doesn’t explicitly mention virtualization. Of course, you do need some smart way to pool your computing resources, and 15,000 VMworld attendees can’t be wrong…right? (Go here for my write-up on last year’s VMworld event.) But, just keep that in mind. There’s more to the story.
In fact, there may be more than one right answer. There isn’t one and only one path to cloud computing. My old BEA cohort Vittorio Viarengo had a piece in Forbes this week talking about virtualization as the pragmatic path to cloud. It can be. I guess it all depends what that path is and where it goes. It just may not be ideally suited for your situation.
On the “path to cloud computing,” to borrow Vittorio’s term, there are two approaches we’ve heard from folks:
Evolution: No, Charles Darwin isn’t really a big cloud computing guru (despite the beard). But many companies are working through a step-by-step evolution to a more dynamic data center infrastructure. They work through consolidation & standardization using virtualization. They then build upon those efforts to optimize compute resources. As they progress, they automate more, and begin to rely on orchestration capabilities. The goal: a cloud-style environment inside their data center, or even one that is a hybrid of public and private. It’s a methodical evolution. This method maps to infrastructure maturity models that folks like Gartner talk about quite a bit.
Revolution: This is not something you studied in history class involving midnight rides and red coats. If organizations have the freedom (or, more likely, the pressure to deliver), they can look at a more holistic cloud platform approach that is more turn-key. It’s faster, and skips or obviates a lot of the steps mentioned in the other approach by addressing the issues in completely different ways. The benefit? You (a service provider or an end user organization) can get a cloud environment up and running in a matter of weeks. The downside? Many of the processes you’re used to will be, well, old school. You have to be OK with that.
Forrester’s James Staten explained ways to deliver internal clouds using either approach in his report about why orgs aren’t ready for internal clouds in the first place. Both the evolutionary and the revolutionary approaches are worthy of more detail in an additional post or two in the near future, I think. But the next logical question – how do you decide what approach to take? – leads to the next bit of useful advice I’ve heard:
When in doubt, pick ‘C’. Even customers picking a more evolutionary approach won’t have the luxury of a paint-by-numbers scenario. Bill Claybrook’s recent in-depth Computerworld article about the bumpy ride that awaits many trying to deliver private clouds underscores this. “Few, if any, companies go through all of the above steps/stages in parallel,” he writes. “In fact, there is no single ‘correct’ way to transition to a private cloud environment from a traditional data center.”
So, the answer may not only be a gradual evolution to cloud by way of increasing steps of virtualization, automation, and orchestration. And it may not only be a full-fledged revolution. Instead, you want to do what’s right for each situation. That means the co-existence of both approaches.
How do you decide? It’s probably a matter of time. Time-to-market, that is. In situations where you have the luxury of a longer, more methodical approach, the evolutionary steps of extending virtualization, automation, and standardization strategies is probably the right way to go. In situations where there is a willingness, eagerness, or, frankly, a need to break some glass to get things done, viva la revolution! (As you probably can guess, the CA 3Tera product falls into this latter category.)
Learn from the past. Where people have gotten stuck with things like virtualization, you’ll need to find ways around it. Sometimes that will be helped by tools from folks like VMware themselves, broader management tools from players like, oh, say CA Technologies or a number of others. Sometimes that help will need to be in the form of experts. As I previously posted, we’ve just brought a few of these experts onboard with the 4Base Technologies acquisition, and I bet there will be a few consulting organizations in the crowd at VMworld. Just a hunch.
Back to Claybrook’s Computerworld article for a final thought: “[O]ne thing is very clear: If your IT organization is not willing to make the full investment for whatever part of its data center is transitioned to a private cloud, it will not have a cloud that exhibits agile provisioning, elasticity and lower costs per application.”
And that’s enough to ruin anyone’s summer vacation. See you at Moscone.
If you are attending VMworld 2010 and are interested in joining the San Francisco Cloud Club members for drinks and an informal get-together on Wednesday evening before INXS, go here to sign up.
Thursday, August 26, 2010
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment