Mike D’s Blog

I’d like to officially add another person to my list of those that “get” cloud. Congratulations to Dave Malcom, CTO of Surgient. I know what you’re going to say, “why are you giving props to the competition?” Well, at this point in cloud computing’s development those who understand what’s going on need to band together to push this concept forward – especially when you have companies like Oracle running in the complete opposite direction of cloud. So what did Dave say that made me perk up and write a post? Simple, he wrote the truth in plain english. Go here to read the post and then come back for comments.

Dave lays out 5 characteristics for cloud computing. Here’s my color commentary on each:

Characteristic 1: Dynamic computing infrastructure

In this section, Dave says that the infrastructure needs to be virtual. Good start. He also says it needs high levels of availability and redundancy and it should be able to move services around to quickly respond to workload demands. Sounds a lot like VMware Fault Tolerance, VMware HA, VMware SRM, and VMware DRS play right up his ally. You can’t get more reliable, available, and respondent than what each of these technologies provides.

Characteristic 2: IT service-centric approach

Here’s where people (mainly the IT workers) start to throw you out the door. With cloud computing you need to transfer control to the application or service owners instead of having control in the datacenter. The owners of the service (often the LOB) understand what they want far better than the guys in the datacenter. (I probably just offended every one of my readers now). It’s the truth though. In order to get to cloud the datacenter people will manage overall capacity and the decision of what services to deploy and to where are handed over to the app owners. After all it’s the service that’s enabling the LOB to make money for the company.

Characteristic 3: Self-service based usage model

Dave says users should have “the ability to upload, build, deploy, schedule, manage, and report on their business services on demand”. I absolutely agree and so will most CTOs and CIOs. The good news is VMware fits into this model as well. You can use something as simple as VMware Lab Manager to provide this front-end portal or something more flexible (and complex) like VMware vCenter Orchestrator to build your own portal. Customers are doing both today.

Characteristic 4: Minimally or self-managed platform

The key to building a real cloud is to have the infrastructure manage itself. What I mean is you simply define a service and tell it what resources, security, etc it needs and then lay it on the cloud layer (at VMware we call this the “cloud OS” – aka vSphere). The cloud OS then talks to the infrastructure and configures it as necessary to meet the service needs. There’s a lot of this integration going on today if you look at the integration between the Cisco Nexus line and vSphere or even the EMC plug-ins and vSphere. More work needs to be done in this area and a TON of standards need to be built out but we’ll get there as an industry.

This is the other time that the datacenter people kick you out of the meeting. If the infrastructure is responding to the applications then it means you don’t need as many people in the IT shop concentrating on this. You don’t need a SAN team of 50 people for example that look at every single I/O request coming from an app to figure out if it needs 13 spindles or 13 1/2 spindles. The application defines this and the infrastructure responds to it as the app grows or shrinks. This means you might only need 4 people on the SAN team to keep up with technology changes and monitor capacity needs. This frees up the rest of the team to work on higher order tasks and strategic projects. Of course a lot of IT people have specialized so much that you’ll either need to retrain some of the people in new IT skills or get rid of them to get people in with broader skill sets (there went the rest of my readership). This can be a major blocker to the adoption of cloud computing and is why it needs to be bought off on from the highest levels of the organization first and then filtered down as a mandate to the rest of the organization.

Characteristic 5: Consumption-based billing

This is sort of the generic definition of cloud – you pay for what you use. Most enterprises will have a hard time with this because very few do chargeback to LOBs today. This should change though with the move to cloud. The good news is VMware is putting the pieces in place to do this. VMware vCenter Chargeback will allow you to generate the information necessary to do successful chargeback to the LOBs based on the consumption of resources.

The Responses

If you look at a lot of the comments they’re all pretty typical. Basically what you have is a bunch of IT datacenter people scared of cloud computing. These are the road blockers for the move to cloud since what they care about and know is about to change in a big way. You’ll see most of the FUD come out of these groups. It’s worth reading the comments though since you’ll also see some of the challenges come out of the crowd. Stuff like security, compliance, visibility, etc. All of these challenges are being worked on and some of them have already been solved. Still, you’ll hear a lot of noise over the coming years from the datacenter centric crowd that still wants to taste and smell their infrastructure on a daily basis. These are some of the same people that said virtualization was a fad and will die off at some point. They’re the people resistant to change. It will be interesting to see where they end up.

Conclusion

Cloud is still changing. I think the industry is quickly converging on a good definition for what it is and what the required pieces are. It’s good to see another company out there with someone smart steering the ship that “gets” cloud. This will help out the industry as a whole a lot.