A couple of months ago, one of the leading x86 server virtualization software vendors (the self-styled “market share leader” in a smallish pool comprising roughly 17 to 20 percent of all installed servers [those actually running server hypervisor software]) held its annual conference in Northern California. At the event, the vendor introduced a modest, if incremental, update to its flagship product that seemed to fall short of evoking much excitement even from the usual crowd of devoted fans, let alone the few hundred members of its “vendor ecosystem,” who occupied the caravan of tents in the middle of the exposition center.
I wondered whether the blush was finally coming off the rose of x86 server virtualization. Some folks I had interviewed recently seemed to say so, including one fellow, formerly the EMEA director for the server virtualization giant of the moment, who told me he had jumped ship a few months before, citing “a dearth of six-figure deals” around the product.
Others offered that the momentum behind server virtualization continued to grow, but under other guises. The marketing of server virtualization technology simply morphs its rhetoric every six months. After about six months of one set of jargon, the marketing mavens of server virtualization vendors re-engineer the pitch lines begetting a new raft of “X as a Service” terminology: Software as a Service, Infrastructure as a Service, Platform as a Service, etc. Six months later, the marketects change terminology again and begin talking about “clouds,” which comprise yet another take on the same basic technology. Now, we’re beginning to hear a lot about “software-defined” XYZ: software-defined networks, software-defined data centers, etc. Same dog, different fleas, so the saying goes.
Anyway, while attending the vendor’s confab, I chatted with many of the folks who were staffing vendor partner booths in search of a modicum of inspiration that might lead to an article (or a column, a blog post or a tweet even). Instead, what I heard over and over from booth operatives was that their product “fixes a problem created by the server hypervisor,” the flagship technology of the show.
That was when it dawned on me that the disruption (and outright destruction) caused by traditional x86 server computing had the advantage of creating opportunities for all these vendors—creating, in effect, a vendor ecosystem based on parasitism.
“Parasitic” may strike some readers as a bit of an overstatement. After all, it isn’t uncommon for ancillary vendors in an ecosystem to “draft” off the big vendor much in the same way the pack follows behind the lead car in a road race, leveraging the willingness of the leader to burn his fuel and resources (marketing budget) to cut through the air ahead. From that perspective, all vendor ecosystems are parasitic to some degree.
The big vendor is OK with this parasitism because it gives the impression that its technology is “game changing” and the entire industry is in lockstep. The reality for the customer is that the new technology breaks things—server hypervisors are playing havoc with existing local area networks (LANs) and storage fabrics in many organizations where they’re deployed—and in so doing require significantly more CAPEX investment than the hypervisor vendor originally suggested.
After you calculate the costs for all the hypervisor licenses, ancillary software licenses, newer server gear tricked out with all the I/O cards that may be required by the most demanding guest application, plus consulting expenses and specialized training and certification for IT staff, you have shelled out only about 20 percent of the overall price for embracing x86 virtualization. Now, according to the ecosystem vendors I interviewed, in order to realize the efficiencies promised by the hypervisor vendor, you also need to re-engineer the storage area network (SAN) you just spent 10 years deploying and the LANs that have grown up around your systems over the years.
In 2005, a leading server virtualization advocate (and “independent” industry analyst) projected that the CAPEX spend for server virtualization would be complete by 2010 and everyone would be reaping in huge CAPEX and OPEX (fewer IT staff) benefits. This has yet to happen. What we have instead is a parasitic ecosystem kept alive by the destructive business model of the leading server virtualization vendor. For the record, I would give zEnterprise another look.