On Wed, Jan 25, 2017 at 12:56 PM, James Wilkins <James.Wilkins@xxxxxxxxxxxxx> wrote: > Apologies if this is documented but I could not find any clear-cut advice > > > > Is it better to have a higher PG count for the metadata pool, or the data > pool of a CephFS filesystem? > > > > If I look at > http://www.slideshare.net/XiaoxiChen3/cephfs-jewel-mds-performance-benchmark > - specfically slide 06 – I can see they used 32.768 for metadata and 8.192 > for data pool. In general your metadata pool with have many fewer objects than your data pool, so you can get away with a lower PG count. If you target SSDs for your metadata pool then you'd definitely want a lower pg num to reflect the (likely lower) number of SSDs. John > > > Cheers, > > > > > > > _______________________________________________ > ceph-users mailing list > ceph-users@xxxxxxxxxxxxxx > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com > _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com