Infoworld has published a quite interesting comparison between Citrix, Microsoft, Red Hat, & VMware at Virtualization shoot-out: Citrix, Microsoft, Red Hat, and VMware. Its quite detailed and the methodologies and comparison basis was documented clearly which make it worth going through.
Although I agree with the comparison to a great degree I still had few comments to post on it. First of all let look at the great news the article has brought:
– VMware are the leaders when it comes to Advanced/Creative features.
– VMware were the only hypervisor to give a consistent performance with every OS & all kind of loads. (Hyper-V performed well when virtualizing Windows, but not Linux or other OS’s. Where RHEV & Citrix Xen performed Well for Linux OS, but not consistently well for Windows). Although it was mentioned in the comparison, I believe this point need to be given a great emphasizes as you don’t want to have a different hypervisor for different OS’s and work load in your environment, when you can have one Hypervisor that will perform will for all. Further, using one hypervisor for all will give you a better utilization of your resources.
– VMware has the widest guest OS Support, over double of the closest hypervisor. Imagine even just testing for a very popular Linux edition Red Hat 6 in this article, they had to use Red Hat 5.5 with Hyper-V as 6 was not supported.
– Something that grabbed my attention was that RHEV has been ranked higher in the review than Hyper-V & Citrix, although I have not been seeing them much in the real world competition. I guess Microsoft Marketing has a much louder speakers than others :).
– First Glitch that got my attention that the comparison has mentioned the gab between the features in all of these hypervisors are getting closer & the underestimation of powerful features that vSphere has & competitors does not. I believe features like Storage VMotion, DRS, vDistribute Switches, VAAI, ability to run Nexus switches virtually, vShield, Network/Storage I/O control, & host profiles are all getting very critical for organization of all sizes not just for the enterprise. Further, small companies should put it strategy that it will grow to an enterprise one day & then it should have the option to pay for these features it need rather than a full rip and replace the full infrastructure when these features are required.
– Second glitch I can see is the way the performance was measured. The performance was measured by a very simplistic loads where SMP performance were not really pushed to the limit. I believe if performance tests would considered Tier one apps where multiprocessing performance to be pushed to the limit performance gap between vSphere & other hypervisors would have been greater.
– Third glitch when pricing comparison was carried out, it was carried out based on cost per physical server, rather than considering cost per VM which seems to be a better measure to document ROI & TCO in Virtualization scenarios.
Over all I believe the article was worth reading & quite fair to all nonetheless of few glitches, but hey who can write for perfection these days.
If you read the comparison and my review above and had a thought, then please share it in the comment area below.