Virtualization is all the rage these days as IT organizations now routinely look for ways to optimize both workstation and server hardware, And nobody has benefited more from that trend than the folks at VMware, which was were scooped up last year by EMC for $635 million.
Back then VMware and a company called Connectix that was acquired by Microsoft were the only two virtual games in town. But a plethora of choices are about to emerge that will significantly alter the virtual landscape.
Most interesting is a startup company led by Scott Davis that should launch in the first quarter. Davis originally worked on VAX/VMS clusters and Windows NT clusters before Digital sold that technology to Microsoft in 1990s. Today, Davis is back with a plan to advance the state of the art of virtualization by allowing any operating system to automatically invoke and add any number of processors on the fly.
That differs significantly from VMware's approach, which allows people to run multiple instances of an operating system on the same machine. VMware is typically deployed on one machine but the company does have and add-on module that will allow people to deploy multiple operating systems on a four-way server.
The knock many people give VMware is that it originally came about as a hack to help developer's maximize their limited hardware resources and as such was not developed with distributed computing requirements in mind.
Dr. Mendel Rosenblum, chief scientist for VMware, argues that performance issues related to running a single instance of an operating system across multiple processors makes any distributed approach to virtual machines extremely difficult. In addition, that implementation would have to be tuned for a specific version of an operating system, while VMware gives people the freedom to deploy any version of Windows or Linux on the same machine.
While future iterations of VMware and rival next generation offerings will surely duke it out in the market, there are several other projects in the works that are likely to have a significant impact in this space. For instance, there are a number of open source virtual software projects, the most promising of which is Xen. That project earlier this month gained the support of Red Hat, Novell and Hewlett-Packard and with their help is expected to be baked enough for enterprise usage in 2005. The support of Red Hat and Novell effectively commits IBM to supporting Xen because IBM resells both Linux distributions. And there in lies a potential conundrum.
When EMC bought VMware, they decided to keep it running as an independent subsidiary largely because most of the people using VMware had been turned on to it by IBM. If EMC had rolled VMware directly into EMC, IBM would have just pulled the plug on support for VMware. This strategic compromise allowed everyone to save face while letting EMC grab a strategic piece of software juts before it was reportedly going to fall into the hand of Veritas. Come 2005, however, IBM may have a number of VMware alternatives that may pave the way for a parting of the ways with VMware and its dreaded EMC parent. And once that happens, VMware's days as an independent subsidiary of EMC could be numbered.
VMware's Rosenblum says that Xen is a different type of virtualization offering that must be tuned for specific operating system implementations, so he expects that IBM will continue to see value in supporting both VMware, which can run under both Windows and Linux, while Xen runs on a specific version of Linux.
While the industry waits to see how all these events will pan out, Intel is working away on projects called Silverdale and Vanderpool that in tandem that at the very least will make virtual software technology more widely available or perhaps even someday eliminate the need for virtual software all together.
It will be 2006 before any of Intel's efforts have any effect on the market and it might be the end the decade before Intel manages to completely embed all the necessary virtualization software required into the firmware of its processors.
Given the amount of interest in virtualization, there is probably any number of additional next generation projects underway. But for the time being at least, the two guys to watch in this space are Scott Davis and Dr. Rosenblum as they represent the virtual vanguard of enterprise computing. And no matter what ultimately happens, this whole are is likely to be one of the most exciting areas of innovation in enterprise computing for years to come.