Am 28.01.2014 17:25, schrieb Peter Matulis: > Within the formula [1] there is an assumption that all pools contain the > same number of objects. That's nearly always not the case. From my experience, the two conditions that should be fulfilled are: 1. There should be a sufficient number of PGs per OSD. 2. The data size distribution of all PGs should be within reasonable bounds. When condition 1 is not fulfilled, the I/O load does not spread evenly through the cluster. When condition 2 is not fulfilled, recovery is not going to be smooth. Ceph warns when the data size of any single PG is more than ten times the average PG size. So there is no hard-and-fast rule for PG sizing. I see some heuristics which should be observed. @Ceph devs - please correct me if I'm wrong. HTH Christian -- Dipl.-Inf. Christian Kauhaus <>< · kc@xxxxxxxxxx · systems administration gocept gmbh & co. kg · Forsterstraße 29 · 06112 Halle (Saale) · Germany http://gocept.com · tel +49 345 219401-11 Python, Pyramid, Plone, Zope · consulting, development, hosting, operations
Attachment:
signature.asc
Description: OpenPGP digital signature
_______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
