Is It Cloudy Up In Here?

I love walking into the data center and hearing all those fans whining, lights flashing, disks spinning. Home and mother. My domain. I can explain how each and every one of them is used – except for a few in the racks over in the corner. You can almost feel the electricity flowing through the chips. In a fair-sized computer room, you can’t hear yourself think – or anybody talking. So there’s some extra benefit to all that hardware.

Go back to the desk and there’s an email saying someone needs a new environment. Hmmm. All my servers are busy, that’s what my spreadsheet says. Well, not so much busy as dedicated to one or two applications at most. Maybe two servers for some applications. Or more. Plus the disk space. And they need this new environment when? Tomorrow? It’ll take that long just to figure out how we can shoehorn this one into our data center. Installing all the software will take another few days. And my staff is slammed. Situation normal. And the CEO is waiting because this is mission critical. This is why you never stopped communing with the silicon even as a manager. Go ahead and jump in!! Super Geek to the keyboard! Saving the day. Or maybe you push back for a few days, buying time to get the time to get the thing installed.

Is that scenario anything you’ve encountered? It happens every day in our world. And it’s happening more often. Your data center is sized for the needs you know about, plus a bit for expansion. But you can’t support a whole new application in Production in a few hours. You don’t have the compute resources or the disk space or the network bandwidth. But this is an irresistible force since the CEO is backing the new application since it means new markets and increased revenues.

Alternate scenario: You need to expand Production but just for a month. I mean you need to add Christmas-shopping scale to the environment, like three times the processor capacity you’ve got in the data center. Buy more hardware? In your dreams. Rent it? No way. Just finding a place to stuff the hardware in your data center would be a major victory. And then you have to test it! Or you could hope it works as expected and doesn’t take you off the air for a day or two in what my wife calls an RGE (resume generating event).

What’s a data center manager to do? Enter the Cloud. We are definitely in the era of the virtualized workload. (I love the concept of workloads, not just applications or servers since it’s really workloads that we’re supporting.) VMWare says that more than 70% of all workloads are virtualized. Those virtualized environments are just perfect for the cloud since a virtual server is pretty much a virtual server no matter where it lives. The same study tells us the normal utilization for physical servers is less than 20%. Talk about wasting resources. Don’t let the CFO know! For all that, if you have LPARs or Zones or PureSystems or Exadata/Exalogic, you have the beginnings of a Private Cloud. You're going to be talking about services that support your workloads. It's all just new names for what you're already doing.

Don’t be put off by the names for the types of clouds. Private, public, or hybrid. Want an easy to remember them? I use this dating analogy. Private = my place. Public = hotel lobby. Hybrid = connecting rooms. Well, that’s my suggestion for what it’s worth.

You’ll need some guidance to ensure that you don’t wander off into some disused siding. Every vendor has a reference architecture to help ensure your Cloud success. Get a Cloud Architect in for a few days to build a custom reference architecture for your Cloud usage. (I’ll bet your hardware or software vendor will send one at no charge if you whine a bit.) What you’ll get is a guide to help you through the various intricacies and nuances of your journey, from provisioning, security and deployment. A flow chart would be great! After all, we don’t read directions, we’re in IT. We look at the pictures!

Why are you devoting capital dollars to development and testing resources? Aren’t these OpEx activities? Cloud is just the thing for moving from CapEx to OpEx. And you have the choice to drop any of your Cloud environments with just a few keystrokes or a phone call when you no longer need it. If it’s OpEx, the cost will go away when the need has been satisfied which means you don’t have those pesky ad hoc environments hanging around.

Finally, there are all these XaaS acronyms flying around. Another reason to stay away from the Cloud – You don’t speak the language. IaaS. PaaS. SaaS. DBaaS. What the …? We’ll decode the Services next time in Defogging Cloud Services

Note: Parts of this article were published previously in the IBM Data Management blog as “Feet in the Data Center, Head in the Cloud”