Recently I wrote an article about the overall “vCloud EcoSystem”, and what components were required to build out a full Cloud environment. The article briefly touches on the availability of some of these components. A quick extract is shown below:
So why are these components important? When designing a vCloud environment, you need to take into consideration the availability of certain components, like vCenter for example. This is no longer a management tool that is used to manage your virtual infrastructure. This is a critical component of your vCloud environment passing through all the operational commands a consumer initiates using vCloud Director. How do you protect vCenter? There are 6 databases in a vCloud Ecosystem. How do you protect all the databases? Every component shown above with the exception of ESXi and vCloud Connector have a database that have relationships with each other. If you loose a vCenter database and you restore a database from two hours ago, how does vCloud Director know what to do now, or Chargeback for example? These are all considerations when carrying out a vCloud design and implementation.
For the Brown bag session that Chris Colotti and I presented for ProfessionalVMware.com. I talk in great detail about these components and there availability, I created a nice diagram showing the different touch points and people have been asking if I could share this out. (Note this is an update of the digram in part 1)
Looking at this diagram, you can see that each component has some kind of touchpoint to not only vCloud Director, but also vCenter. This is where you need to think about synchronising backups when performing your vCloud Design. But it can get a lot more complex, what happens if you start building out your vCloud environment further. vCloud Director supports up to 25 vCenters. How does this effect the diagram?
Now with this example, you have more of the components with multiple touch points into multiple vCenters, all providing services to your vCloud environment. Now if we were to expand this even further, and add more vCenters, how does this affect the environment?
There is now multiple touch points, that require not only backing up, but need to be highly available to provide specific SLA’s. From a service provider perspective, if you loose a vCenter you have lost the cloud services you are providing for the pVDC’s running within there. vCloud Director will point this out to you too, which is not a great idea for anyone who is charging for a service. Imagine you are the consumer, and have just paid £xxx for some resources and you are told they are unavailable. This would not be a happy conversation I would have with a provider, thats for sure.
So to recap on both of these articles, when designing your whole vCloud infrastructure, think about the multiple components that make up a vCloud environment, the SLA you want to be able to deliver, and consider how you grow this environment when you have more demand. The main key principal to cloud computing is to be able to scale out when demands require it. Don’t build your vCloud environment totally insular to one vCenter, you may need more one day.
If you get chance, I would recommend listening to the Brown Bag that Chris and I presented. You can watch the video’s by clicking here.