Yesterday I spent half a day discussing virtualization strategies with a few CTOs from a $15B firm from the healthcare industry. It was refreshing to see the same glint in the eyes of these CTOs regarding virtualization’s potential to revolutionize their data centers as I saw before joining VMware. The key aspects of these discussions was how virtualization can help them reduce not only their capex but also their opex. They also saw the potential that virtualization provides in the areas of increasing their service delivery times (provisioning new systems for customers in 2 hours instead of 2 weeks or 2 months like they are used to) as well as enabeling a new way of thinking about disaster recovery.
Like most IT executives, these CTOs knew about virtualization from the aspect of the hyperadvisor; the core of virtualization that allows you to run multiple operating systems on the same computer at the same time (with each operating system thinking it’s by itself). This is just the start of the software computer revolution. Now that you have isolated the OS down to just a bunch of files that can be moved between physical computers transaprently, you have the ability to provision a machine in a short period of time. Create a base Virtual Machine image for each of your approved OS’. Next copy that base OS to create another image for each of your general application types. You now have a library of images that can copied within seconds to an existing physical machine, booted and viola, a provisioned server in minutes; substantially reducing opex.
Also, since the “data” that the enterprise is already backing up for disaster recovery now contains the applications and operating system, DR takes on a whole new meaning. (and not just DR, but resource “load balancing”, server maintenance, data center migrations…pretty much all the maintenance work that IT has to do.) You can take your old machines and deploy them to your DR site. Now use your existing data replication software to replicate the VM files. When your main data center goes down, just turn on your VMs in the DR data center and back to business. Sure, performance might not be the same, but the business is functioning; the key point of DR. And since you used hardware that you already own, your capex is reduced.
“Cloud Computing for the Enterprise”. That is what one of the CTO’s called virtualization. As someone who has been living in the cloud for years on the Web 2.0, this is not a shocking way for me to think. Cloud computing was easy for the web, where you had (or could build) apps out of standard web based (REST based) building blocks. But for an enterprise architect who has to deal with complicated software packages from properitary vendors that process Billions of dollars in orders, this was a refreshing thing to hear.
Could this type of statement indicate that virtualization and utility computing has arrived for the enterprise?
I think that’s a safe statement.