Distributed Storage

[finish06]

Hello!

Setup: 3 node Proxmox cluster with x2 500 GB SSDs per system at the moment. Currently using external NFS storage to allow for migration and HA, however interested in decreasing dependency on external NAS.

Question: Would I be best to use DRBD or ceph to allow for easy migration between host? Can I set up DRBD above a RAID10 array (I am looking to increase drive number to four within coming months). I really want to speed increase associated with RAID10... is this possible with Ceph? However, how would Ceph rw speeds be hurt by the small number of disk?

TL DR - Can DRBD or ceph be placed/above/on top of a zfs raid 1/0?

 

[spirit]

Go for ceph, from my opinion, drbd is not enough stable (and limited to 2 nodes anyway with drbd).

With ceph you can scale-out to any number of nodes, and grow/shrink the cluster in time.
you need plain disk for ceph. (no need to have raid, ceph manage the replication)

ceph is not hurt by small number of disk, not anymore.
you just need to be carefull, that your ssd can do fast sync writes for journal (same for zfs by the way)
http://www.sebastien-han.fr/blog/20...-if-your-ssd-is-suitable-as-a-journal-device/

 

[paulocoghi]

Is it possible to run OpenVZ and/or LXC containers directly on Ceph? Or another distributed file system?

 

[tom]

Is it possible to run OpenVZ and/or LXC containers directly on Ceph? Or another distributed file system?

yes for LXC.