On 23 Apr 2013, at 17:06, Sylvain Munaut <s.munaut@xxxxxxxxxxxxxxxxxxxx> wrote: > Hi, > >> - client dom0: a simple quick debootstrap, and a low amount of memory to bypass buffers > > I assume you meant domU ? > You ran those tests in a VM right ? Right! DomU indeed, not dom0. The benchmarks are done on /dev/xvdb, being the rbd device. I was able to block-attach/detach them also on the dom0, but that is trivial once it works. > >> Thx for the work you've put in this! Seems to work nicely at first glance, but I'll do some more tests later on (with a more recent ceph cluster)... > > Thanks for testing. When you do more tests, it'd be interesting to > compare the kernel driver with the blktap driver. I'll be upgrading my testcluster first to the latest stable ceph release, and then run some endurance-tests against the setup. But that won't happen before mid next week. > > I've actually identified a bottleneck caused the Xen IO ring splitting > large requests into small 44k chunks, which tends to lower RBD > performance a lot ... I'l still investigating possible solutions to > that. Feel free to bug me once it is fixed, if you want some testing. Rgds, Bernard Openminds BVBA > > Cheers, > > Sylvain > -- > To unsubscribe from this list: send the line "unsubscribe ceph-devel" in > the body of a message to majordomo@xxxxxxxxxxxxxxx > More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
