Another view of virtualization

I’ve spent the large part of the last five years looking at server and desktop virtualization. It was only a matter of time before I was going to spend cycles in this blog focused on the role, impact and future role of virtualization on datacenter IT and ultimately on the desktop. I will likely continue coming back to this subject or one of its derivations (Cloud Computing, SaaS, etc.) over the coming months.

I thought I’d start with my view of the hypervisor on IT. The hypervisor is a disaggregation technology: it disaggregates hardware from software, x86 platforms from operating systems. For those Macintosh or Linux fans out there, the hypervisor is what easily brings Windows applications to your beloved platform. For IT the hypervisor allows applications to continue to run as you migrate to new platforms or seamlessly add additional workloads to existing platforms.

My contention has long been that for the obvious utilities -- consolidation, optimization, business continuity solutions -- the hypervisor has real value but only evolutionary value. Best practices don’t fundamentally change. Architectures evolve but don’t revolutionarily change. Management architectures do, even sometimes radically change, but the tools of the trade don’t really do more than evolve.

During my tenure at VMware I was struck by the fact that there are areas (two distinctly) that are revolutionarily impacted by the hypervisor. Two areas that, at the time, VMware wasn’t only minimally invested in. These two areas -- virtual appliances and desktop virtualization -- have become, under new leadership, more highly invested in at VMware but still, I believe, entirely under invested by IT.

The first area, virtual appliances, seems to be the least impactful over the last few years. There seem to be only a few companies trying to create a business around this concept. Virtual appliances, or the use of a hypervisor by a software developer to disaggregate OS decisions for their customers, could potentially have a profound impact on software development models and dramatically change the call-to-arms between warring .NET and J2EE camps. The ultimate impact of a virtual appliance was captured, albeit only briefly, by BEA (since acquired by Oracle). Since BEA had a fast derivation of an OS (their implementation of JRocket JVM) they wedded ESX with an optimized JRocket and their app server to create a “bare metal” implementation of their software stack for Intel-based systems. No need for Windows, Linux, Solaris or any other OS. Simply install the stack with a single click and you have an optimized software solution installed and ready-to-run. BEA could pick and choose which OS-level components it wanted to optimize and deliver as part of the stack. While the utility is obvious for the customer the ultimate savings were never realized by BEA -- no longer having to choose a single or multiple development paths for different OS platforms -- as they were shortly thereafter acquired.

There are some companies out there trying to build a software development business around this technology. rPath is one that comes to mind. It’s also clear that VMware has increasingly invested in software distribution for their partners using the hypervisor in this manner. They have built a software distribution community through their corporate site focused on this. However the long term impact on the “traditional” developer model hasn’t yet developed. I have faith that at some point it will. Perhaps it needs a bit more consumer exposure?

The other area, desktop virtualization, has been an area of heavy VMware investment over the last couple of years. It’s also an area that I know quite a bit more about having spent a year in thin clients with Sun Microsystems and the last couple of years with a desktop virtualization start up (which I am no longer associated with). I have also done quite a bit of writing about this space (and will obviously continue to) -- see my very first blog entry. Stay tuned for tomorrow’s update for more on the evolution of the desktop and the impact of the hypervisor on OS delivery.