I went over to the InfoWorld Virtualization Forum this week to get a better understanding about the technology behind the buzz. I figured I was missing something when I read all the hype that built up around the VMware IPO and other virtualization concepts. But when I heard Stephen Hilton, an IT executive at Credit Suisse (which is doing a big server virtualization project) say virtualization â€œwould drop out of the vocabulary in three years,â€ I felt better.
I was convinced before I went that â€œvirtualizationâ€ is just a buzzword du jour, today’s â€œdot.com.â€ So I was predisposed to hear what I wanted to hear Iâ€™m sure. But I understood his point to be that this was nothing new and just the way IT infrastructure needs to be built. Thatâ€™s the way I understood it too. IBM has had it for 40-50 years on mainframes; it was only a matter of time before it would be used to better utilize PCs on networks. Eventually, it will be built into every type of server software; freestanding virtualization will go the route of the spell checker, from separate product to a check-box feature of a higher order product.
To be clear, I understood the concept vis-Ã -vis servers going into the conference venue. Working on Multics marketing in the early 1970s, I was there at the beginnings of the virtualization that EMCâ€™s SEC filing treats as if it was ancient history (which I guess it is). The logic says â€œLetâ€™s do for the 25 million x86 servers out there what VMS and AOS/VS did for VAX and Eclipse in the 1970s.â€ And what IBMâ€™s VM did for its proprietary systems at the same time Multics was commercially announced. Given the 30-year-plus maturity of this concept and its inevitable commoditization, the barrier to market entry is low. VMware has a first mover advantage (usually short-lived in IT marketing) but Red Hat, Microsoft, Novell, IBM and others are right behind.
â€œServer virtualizationâ€ adoption uptake will be an interesting process to watch but I donâ€™t see the technology as too exciting for investors long term as the spell checker syndrome kicks in. Not surprisingly, panelists at the InfoWorld event mentioned that cultural issues, lack of standardization of operating systems, shared services issues, and lack of applications that take advantage of server virtualization are bigger issues than the technology itself in terms of adoption. On the other hand, they see virtualization technology as a great way to inculcate some much needed high availability (HA)/failover capabilities into their infrastructures. In fact, I heard more about HA as a benefit than better PC resource utilization in the various user and vendor presentations.
So as for server virtualization: great concept, not a differentiator for very long, kind of like rack and pinion steering and electronic ignition (no other choice after a short time lag).
The other problem is that given that virtualization is such a hot buzzword, it is being thrown into the IT investment research mix in ways that make entirely no sense at all technically. Back when the VMware IPO came out, I said I donâ€™t understand why I would want to virtualize my desktop PC and laptop? According to IDC (according to EMC), less than 1% of non-server personal computer devices are virtualized. I was surprised that itâ€™s that high.
So I was happy to hear that â€œdesktop virtualizationâ€ suppliers donâ€™t want me to virtualize my PC. They want me to get rid of it. They will replace it with X-Windows. Thatâ€™s right: X-Windows. See the series of white paper and case studies I did for HP in 1992. Dave Cohen of Merrill Lynch scoped it all out at the InfoWorld meeting. His caution though was as follows: â€œDonâ€™t just move all those management problems into the data center.â€ Now that I understand what they are talking about when they say â€œdesktop virtualization,â€ I do actually think there is a big market here: for the management of diskless appliances, all the devices â€œout on the edgeâ€ that are not on the desktop. Letâ€™s call it â€œnon-desktop virtualization.â€
And there was another up and coming buzzword, â€œapplication integration.â€ Read that one as electronic software distribution and associated features, also circa 1992.