On 08/30/2017 12:35 PM, Ulrich Windl wrote: >>>> Ferenc Wágner <wf...@niif.hu> schrieb am 30.08.2017 um 10:51 in Nachricht > <87h8wpa16c.fsf...@lant.ki.iif.hu>: > > [...] >> Do you mean the LVM metadata read latency as seen by the LVM tools or >> that of a mirrored data region? > [...] > > Actually I monitor all of these: > (on the VM host) > * the Multpath devices for the SAN storage that provide the PVs > * the mirrored LV that provides VM storage > * the mirror logs of the LV > * the OCFS2 filesystem that hosts the VM files and devices > (on the VM guest) > * the virtual physical disk" > * the fileystem on top of it > > An amazing side note: For comparison I monitor NFS latency in a VM-based > NFS-client where the server is physical, and the delays are a lot more > constant > than for the virtual local disks (see attachment).
Which would mean that at least part of the latency is due to basically unnecessary synchronization effects. Given that the history of making everything asynchronous and decoupled is much longer for networking than for block/filesystem- layers ... Regards, Klaus > > Regards, > Ulrich > > > > _______________________________________________ > Users mailing list: Users@clusterlabs.org > http://lists.clusterlabs.org/mailman/listinfo/users > > Project Home: http://www.clusterlabs.org > Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf > Bugs: http://bugs.clusterlabs.org
_______________________________________________ Users mailing list: Users@clusterlabs.org http://lists.clusterlabs.org/mailman/listinfo/users Project Home: http://www.clusterlabs.org Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf Bugs: http://bugs.clusterlabs.org