Discussion with Bryan Doerr of Savvis

I recently spoke with Bryan Doerr, CTO of Savvis, about VMware's vSphere, Savvis' customers use of virtualization, can a industry standard system-only product legitimately be considered a "datacenter operating system" and life in general.  Bryan has had the misfortue to speak with me in the past and, I'm sure, must have been pressed into doing it once again by VMware's talented PR folks.

A bit about Savvis

Savvis has been offering its customers offerings that would be considered forms of cloud computing since 2004. Bryan would point out that while VMware's products have been favored by Savvis' clients, this concept really requires a broader base of technology that just deploying virtual machines both on local and remote systems. Bryan believes that this concept really requires a sophisticated management environment, security and both network and storage virtualization technology to be really effective.

Savvis and vSphere

Bryan indicated that he's a strong proponent of VMware's vSphere because it provides needed tools when organizations move towards either a in-house or external cloud implementation.

Datacenter operating system?

When asked if an industry standard system-only product could really claim the title of "datacenter operating system," Brian responded that this addresses the requirements of the majority of Savvis' customers. Since Savvis is largely in the business of hosting applications, platforms and infrastructure based upon industry standard systems, it is not at all surprising that he would say that. When I pressed a little harder on this topic, Bryan said that when a customer asks for a complete multi-vendor, multi-platform solution, Savvis would work with its partners to host mainframe and midrange system workloads.  I believe that this was his way of saying, "well, no, mainframes and midrange systems are still important."

Architects should include virtualization in their plans

Our conversation then went on a wonderful exploration of virtualization and I think we would both agree that this concept is based upon an environment in which the capacity of available systems exceeds that needed by an organization's workloads. At that point, it is possible to use that excess capacity to create a highly flexible, highly agile environment. Virtual machine hypervisors, application virtualization frameworks and other forms of virtualization use that excess capacity to create their illusionary worlds.

Bryan would state that a mature design practice would be to move from monolythic to distributed application architectures. This approach would allow architects to make the most use of parallelism and allow applications to offer higher levels of performance, scaleability and reliability.

Bryan Doerr's advice

When I asked what type of advice he would offer others, Bryan suggested that organizations include virtualization when developing new workloads. This, he would point out, can drive efficiencies, reduce overall costs and offer a number of much needed benefits.

Thanks for taking the time to speak with me, Bryan.