By Trevor Connors published: Wednesday, April 04 2007
Simon Crosby's CTO of XenSource comments on this comparison
Recently I have seen several postings about a German paper that I understand to be a performance comparison between OpenVZ and Xen. I however can not be sure due to the fact that the title page is one of the 86 pages that is not present in the abridged English translation.
As I attempted to locate a full translation, I found this posting by Kir Kolyshkin on an OpenVZ Blog. Mr. Kolyshkin makes several good points in his post. He states that "VMware did a not-quite-good job comparing their ESX to Xen" and "my suggestion is to take those benchmarks and comparisons with a grain of salt." Unfortunately, there is no mention of translating the other 86 pages.
At this point I contacted XenSource to inquire if they have seen this "comparison" and what their thoughts were on the subject. This is the reply I received from Simon Crosby CTO of XenSource
RE: Xen vs OpenVZ comparison
OpenVZ solves a different problem than we do. It splits a single linux into multiple virtual application hosting environments, each of which has to run on the same Linux OS. No concept of OpenVZ running Windows Server 2003 SMP guests. So, the only thing one could compare is a particular Linux app set (eg apache) on a particular OpenVZ linux kernel, vs that same app in a linux VM, on Xen. They would beat us on memory use overall, and on raw I/O they'd come pretty close. But the key limitation is all their "virtual jails" are for the same guest OS.
Sounds like comparing apples to pineapples. I guess a full translation will not be needed.
