One never likes to toot one's own horn. But, the good people of BMC Communities put together this podcast and interviewed me on Cloud Computing last week, so I thought it might be of interest to the readers of this blog. If it gets good response, there will probably be more!
As organizations begin incorporating aspects of cloud computing into their infrastructure, they often run into significant challenges, driven by the increased scale and dynamicism of the cloud environment. What they typically find is that their current IT processes, tools, and policies are simply unsuited to the new environment, and that scaling linearly (for example, by hiring additional staff) is both economically infeasible and insufficient – since a cloud environment typically scales exponentially.
The conclusion that most organizations reach is that their cloud computing initiative has a side effect of forcing the organization to improve its IT maturity – to develop automated, integrated, and well-structured processes, policies, and workflows. It’s often this change that proves to be most challenging to organizations – more so than the technical challenges around deploying new tools. But the payoff is doubly beneficial – not only does the organization achieve a Cloud environment, but they also improve the efficiency and reliability of their existing IT environment.
I’ve recently come across several articles and blog entries, where writers talk about “running applications in the cloud”, and then somehow conflate this with other aspects of Cloud computing. Given all the uncertainty and hype around Cloud, this isn’t surprising…I was recently in a meeting where a product manager and an architect got into a shouting match about whether or not SaaS was Cloud! As is typical, each had a different definition of “Cloud” in mind, which precipitated the heated discussion.
Along these lines, I’d argue that deploying and running an enterprise application in the cloud does not necessarily mean it’s a Cloud Service. Taking an existing, on-premise enterprise IT service (such as email), and re-hosting it off-premise does allow you to legitimately state that this application “runs in the cloud”, and can potentially provide a number of financial and business advantages. But, this does not make it a cloud service, in the sense that (for example) it shows up in a service catalog, for rapid provisioning.
Now, of course, if you’re a Managed Service Provider, and you offer hosted email as a service, you would likely have email show up in your service catalog, and be able to be automatically and rapidly provisioned on demand….in which case it could legitimately be termed a cloud service. It all depends on your perspective!
Every so often, you call the airline to change your flight and the nice – or not so nice – person on the other end of the line tells you that the computer is down. Dead. They are unable to change your flight because something is wrong with the mechanism by which they would make the change. And the nerd in you thinks “reboot a little? Re-attach your network cable?” – but of course, the only polite thing to say is “Those darned new fangled machines. I liked life better when we booked tickets using forms in triplicate.”
We really don’t want to know how airline tickets are booked. There’s some crazy big scary system back there that somehow ensures that I never get an aisle seat – and I accept the black box of it all. We also probably don’t want to know how roaming works between wireless carriers or what backwards mechanism keeps the supermarket self-checkout scale accurately identifying that I never put the butternut squash in the bag. But, all these technologies occasionally break down, making us confront, head one, the sausage of the IT world. It’s not always pretty, even for a nerd. There’s usually more to it than we think, and usually more pain involved in fixing it.
The response is ubiquitous, despite our appreciation for complex technology: Just fix it. I paid for it to work. It shouldn’t be my problem.
Public cloud is another one of those things, like wireless towers and cable TV, where it just has to work. As a consumer, you don’t want to know how, where, or why. You never want to hear the words “oh, the vMotion sent it to Spain.”You should never know that it’s going through a platform upgrade. You don’t want to hear “I think it lost an IP address.”None of these things are your problem. It just should work – like a big, delegated black box of Not My Problem.
That’s the value prop. Of course, from the provider side – it better work! You’rebeing paid to take on the headache – and do it reliably. We often think of Public/Private cloud as a CapEx/OpEx distinction. I’d argue it’s as much a transfer of Advil as a transfer of servers, when it’s done right.
The old Girl Scout song was playing in my head this morning when I was considering a recent customer meeting I had. We usually think about “old” in the datacenter as legacy applications and mainframes. Increasingly, there are some flavors of Unix thrown into the mix. But, it was new to think of a VMware x86 environment as “old.”
Of course, it isn’t old. But, when you consider putting up a cloud environment, there is a real and important question of how you get all that virtualized stuff into the cloud. Should be one-click, right? The cloud is just a new view of the virtualized environment. And goodness knows, the devil’s advocates out there will profess that cloud = virtualization. So, why is there even a problem here?
Well, some cloud systems thrive on homogeneity. Net new services all run on one of a few possible configurations and OS’s. Clean. It might force everyone to comply with these rules, and that’s a whole separate blog post – but it also creates the “legacy virtualized” problem. What to do with VMs that don’t comply with the homogenous rules of the brave new cloud?
BMC’s solution doesn’t have any expectation of homogeneity –in fact we pride ourselves on the varied provisioning options we support. So, it never occurred to me that there might be additional fallout from a homogenous approach. To us, we’re happy to include any of those existing VMs as cloud services within our self-service portal, and have them managed alongside “new cloud services.” In fact, if you want… one can be silver, and the other, gold.