P2V - Solaris x86 and Linux workloads

We virtualised more non-Windows physical servers last week - the more Unix/Linux hosts we virtualise, the more the strategy is reinforced; there is no straightforward process that works for every flavour of Linux/Unix currently supported by VMware as a Guest OS.

For example, a while back we P2V'd some Scientific Linux servers, and despite not being officially named on the 'Guest OS' guide, they are basically Red Hat - they even look like Red Hat, the servers run fine. How we P2Vd them, for one reason or another was not straight forward.

A few points to take on board before you jump straight in: 1) there is no tool (that we can find) that is a click and go for these P2V's - some tools free and comercial, work for some variants, but none work for all. 2) Understand the pre-reqs; disk space etc and you'll make an estimate of how long to complete the P2V - then at least double the time it would take to P2V a comparable Windows workload. 3) Make sure you identify the OS release carefully; there are so many Unix/Linux variants, easy with so many patch levels and you need to be P2V'ing a supported release.

Even converting Windows workloads using the 'click and go' method does not always work first time, once its going it generally completes and works like a dream. No such dream to capture the multiple Linux and Unix variants which make up the supported Gues OS pool on ESX 3.5.x.

I am currently writing a brief paper on the process to P2V a Unix/Linux workload; I'll post it up when its finished, essentially it is the best practice as far as we are concerned on methods to try, starting with the easy one - the commercial tools. Why reinvent the wheel.

Anyway, we have so far virtualised (P2V'd) a number of Linux/Unix servers; Scientific Linux, Red Hat; Ubuntu and Sun Solaris 10. I am sure we'll identify more as we go along...