IBM Publishes Their First VMmark Result

Our partners at IBM have published their first VMmark result on VMware’s VMmark results page. IBM achieved an excellent result using their 16-core IBM System x3850 M2 and VMware ESX Server 3.5. Please check out the full disclosure for details. I am delighted that with IBM’s publication we now have four major hardware vendors represented on the results page. I am looking forward to working with them (and others) to generate more results not only to help our customers evaluate the trade-offs between varying virtualization platforms but also to help us build better virtualization solutions.

In other VMmark news, astute readers may notice that we have added a VMmark 1.0 version number to the results summary table. We are in the final stages of putting together VMmark version 1.1 for release. As you know VMmark 1.0 uses 32-bit workloads only. That was a good choice when the benchmark was initially defined. However, 64-bit applications and OSes are becoming more prevalant and we would like the benchmark to remain representative of a typical datacenter. To that end, VMmark 1.1 will comprise a mix a 32-bit and 64-bit workloads. We will have more to share shortly.


One comment has been added so far

  1. One of the things I’d like to see in VMmark is not just the maximum throughput (which is, obviously, the main point of the test) but some idea as to at what point transaction latency starts to suffer.
    It’s handy for comparing systems to see the max results, but it would be lovely to see the results as it ramps up to the full # of tiles so as to have a good idea as to whether we ought to start worrying at 50% utilization, or 70%, or 90%, about performance of individual virtuals.

Leave a Reply

Your email address will not be published. Required fields are marked *