My recommendation would be to for sure use battery backed up raid controllers for spinning disks -just a RAID 0 per disk- and use IT mode controllers for SSDs. We did plenty of testing and spinning disks REALLY benefit from raid controller write cache -in write
back mode-, but SSDs are faster with IT mode controllers.
Saludos Cordiales,
Xavier Trilla P.
Clouding.io
¿Un Servidor Cloud con SSDs, redundado
y disponible en menos de 30 segundos?
¡Pruébalo ahora en Clouding.io!
El 31 des 2018, a les 17:15, Marc Schöchlin < ms@xxxxxxxxxx> va escriure:
Hi,
our dell servers contain "PERC H730P Mini" raid controllers with 2GB battery backed cache memory.
All of our ceph osd disks (typically 12 * 8GB spinners or 16 * 1-2 TBssds per node) are used directly without using the raid functionality.
We deactivated the cache of the controller for the osd disks completely for spinners and ssds in the lifecycle controller:
- writing: to prevent cache access contention (throughput of cache might create latency spikes because the disks do 90% write load)
- reading: ceph is used as storage system for virtualized systems (XEN with rbd-nbd)
-> we already have several levels of caching which should prevent re-reading of data: pagecache of the virtualized systems, rbd cache of rbd-bd, bluestore cache on the osd
Do you see scenarios where it might be a good idea to activate the cache anyway?
Regards
Marc
Am 20.11.18 um 15:11 schrieb Ashley Merrick:
Me and quite a few others have had high random latency issues with disk cache enabled.
,Ash
On
_______________________________________________
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
