On Sunday, 11/18/2007 at 01:29 EST, Pieter Harder <[EMAIL PROTECTED]> wrote: > I have recently implemented Link Aggregation on my Vswitch. I was expecting to > see increased throughput on single Vnic's, but no such thing. Apparently there > is another bottleneck that I can't pinpoint yet. I have been looking at it with > IBM PTK but nothing obvious shows. I do have Velocity ESALPS, but they don't > use domain 8 monitor records yet (eh, Barton, how about it?), so the picture is > sketchy there. Then I went hunting for performance documents in this area. > There are tons of documents focussing on the capacity of various OSA cards, > Hipersockets etc, but none that I can find on what a single Vnic can do. Being > a total software construct I would assume the only bottleneck there would be > the availability of Vcpu power. True or false? Who can say anything on this?
It has always been the case that guest I/O is ultimately gated by vCPU. If it isn't being dispatched, it can't start any I/Os. The place to start is the z/VM 5.3 Performance Report. Link aggregation results are at http://www.vm.ibm.com/perf/reports/zvm/html/linkagg.html Alan Altmark z/VM Development IBM Endicott ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
