In February 2008, the independent company Principled Technologies published an analysis committed by IBM to compare VMware ESX 3.5 performance on IBM and HP quad-core servers: an IBM System x3850 M2 and a HP ProLiant DL580 G5.
As always happens in these cases, the winner is the client:
- The IBM machine produced 27% better performance per Watt than a similarly configured HP machine with redundant power supplies active at five CSUs
- The IBM machine delivered 8% more performance running vConsolidate with the optimum number of CSUs (five) than the HP machine
- With the redundant power supplies active at five CSUs, the IBM machine used 15.1% less power than the HP machine.
The results are anyway interesting because Principled Technologies used the Intel vConsolidate benchmark framework to run its tests.
Somebody may want to try some comparisons with the results provided by VMmark, the VMware benchmark framework, for the same machines but it’s worth to note a couple of major differences:
- the hardware configuration tested with vConsolidate is not necessarily equal to the one tested with VMmark
- in the case of HP ProLiant DL580 G5, the VMmark benchmark was performed with VMware ESX 3.0.2 instead of 3.5.0