VMware Inc., the leading server virtualization provider, has made a lot of noise about cloud computing these days, but the core of the company's business remains vSphere and vCenter, its virtualization suite and management platform, respectively. News Director Alex Barrett sat down with VMware CTO Steve Herrod to discuss vSphere's future of the virtualization platform, vSphere features that deserve a second look, and trends in the computing industry.
In a second portion of this interview, Herrod talked further about VMware's forays into application development in the cloud.
What can customers expect from the next major release of vSphere?
The next major release of ESX will focus on efficiency and control.
On the efficiency side, one upcoming thing from us will be memory compression. Memory has been one of the things that has been left behind [in our development efforts]. By compressing memory, you can get more VMs [virtual machines] on a host, thus greater efficiencies.
On the control side, we are working on an I/O feature, where you'll assign a portion of a server's I/O to a specific VM. That's really important to people putting high-end applications in virtual machines, especially as we move to the cloud, where you need to be able to give performance guarantees.
Consider VMware's existing product portfolio, what strikes you as underappreciated?
VMware Chargeback. It's one of my favorite products for the way it gives exact visibility into how much resources every virtual machine is using, and creates reports that you can bring to the business unit.
The first time people get that report, they say, 'There's no way I used that.' But after that first stage of grief comes acceptance, and they start to find ways to cut back, turn off those 10 servers that were just sucking up power. So Chargeback isn't the deepest product in the world, but it does bring a deeper level of visibility into a virtual environment.
Do you see demand for virtualized workloads across more than one physical box?
A lot of people have worked on that problem: for example, Virtual Iron, and we actually hired one of their co-founders [Scott Davis, CTO of VMware's desktop business unit]. But generally, with multicore processors, it's really hard for one application to take advantage of all of that. There might be a few applications, but I see that number shrinking based on how fast [technology is evolving].
What do you think of the idea of IT as a service provider?
Some of the biggest customers are already starting to look a lot like service providers. For example, Wall Street firms have internal customers that want to be as separate as they can be, so they already have strong multi-tenancy requirements. They're also starting to charge by usage.
CIOs ask about this a lot, and they see it as both a blessing and a curse. On the one hand, cloud computing is a real threat to how they're doing their business, and at the same time, it becomes a way for them to go from where people are cursing IT to where they are impressed with IT and how quickly they respond.
Hindsight is 20/20, but looking back, what could VMware have done differently?
We pretty aggressively cannabalized ourselves -- gave away products -- but I think we could have done that even more aggressively. It's good for broad market adoption, because once people get a taste, they usually want more.
I also think we could have moved faster in the Mac market. We managed to catch up, but we could have started earlier.