IBM, HP demonstrate fidelity problem with 'open' virtualization; KVM? Really?
Old-school systems makers will back anything that's not VMware
IBM, HP, Intel and a host of smallish Linux vendors have launched a brave new group called the Open Virtualization Alliance dedicated to creating an open standard in server virtualization for the enterprise.
The OVA seems to be made up of two main groups, neither one of which is really interested in the purpose for which OVA was ostensibly formed.
The first is Red Hat, Novell and Eucalyptus Systems – Linux vendors transparently hoping a big consortium will help expand the Linux-specific virtualization market enough to make them popular again.
The other faction is made up of companies with much broader product lines that worry virtualization is to the 21st century what PC computing was to the 20th, and that the vendor whose hypervisor becomes a de facto standard (VMware) will be this century's Microsoft (when it was still, you know, <em>Microsoft</em>).
These guys, as ITWorld colleague Brian Profitt concludes, are mainly interested in shoving VMware, Citrix and Microsoft hard enough to make a little space for themselves in the market for the basic computing platform for the next few decades.
Not what I'd call <em>virtuous</em> motivations; not the mom-and-apple-pie virtues implied by the phrase "open alliance," anyway.
The other big question, at least for IBM, HP and Intel, is what happened to the rest of their commitments to "open alliances" for the DMTF's OVF, CentOS, Xen , not to mention proprietary hypervisors IBM has supported to one degree or another, including its own PowerVM, Oracle VM Server, Microsoft's Hyper-V and VMware's proprietary-turned-open-source specs VMI and VMDK or HP's HP-UX-only Integrity Virtual Machines.
IBM in particular has been trying to get a handle on virtualization ever since the mid-'90s, when VMware revived a technique IBM had used very effectively on mainframes and Unix machines, but either hadn't thought to promote on x86 servers, or was simply beaten to the punch by VMware. And Citrix. And Microsfot. And Oracle. And a lot of other companies.
IBM tried to slow and divert the cloud-computing market a couple of years ago as well, according to James Staten, virtualization and cloud analyst at Forrester.
The Cloud Manifesto published by an IBM-dominated group of vendors in 2009 was less an effort to focus development of cloud computing than an attempt to distract developers into discussions of what it was supposed to be for long enough for IBM to take the lead away from vendors that had already outrun it in the market for virtualization and weren't slowing down in the race for cloud.
IBM "got laughed out of the market" for that one, Staten said.
Most likely no one will laugh at IBM for backing such a universally well known and appreciated approach to virtualization as KVM on CentOS and Suse, of course. Most will probably just stand quietly and envy it for being able to associate itself so closely with the sexy celebrity kernel-level virtual-machine implementation of the moment.
And the OVA will set a new standard for openness and effective integration of virtualization technologies, not to mention making sure there was at least one hypervisor consistently available from reliable, mainline old-school vendors for x86 systems – not like fly-by-night versions like Xen, XenServer, Hyper-V, OracleVM, vSphere, VLX, Trango, Sun VirtualBox, Hitachi Virtage, Microsoft Virtual Server, Parallels...
Well, I think there are probably a couple more as well. Not from IBM, though (except a few of them).
Thankfully, by signing on with another virtualization standards organization, IBM and its partners in this round of Virtual Follies have cleared up the picture quite a lot.