Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
On March 13, 2009, Daniel Bourque wrote: > Hi, > > I'm wondering which way gives the best performance and reliability: > > 1) create a DRBD device on a raw partition, then use it as a PV > 2) use a raw partition for a PV, then create a DRBD device on a LV if you throw in Xen into this heap - you can build your own sandwitch of technologies and it's only up to you what is the best combination for your usecase. LVM gives you online resize, data migration and snapshotting, DRBD gives you redundancy. So if you have fixed device size and use it all for DRBD and then stack LVM on top of it - you'll maximize potential of both. However if you want to take snapshots of entire environment (depends on types of snapshots etc.) you might consider stacking DRBD on top of LVM. With Xen in the mix you might create LVM LV, stack DRBD on top of it, use DRBD as a raw device for VM and use LVM inside of VM. We haven't found in our environment much difference in performance, however we felt that certain features were more important to us so we went with combination that provided it. LVM+DRBD LV can be resized online (followed by online resize of DRBD) with resize2fs. You can take local snapshots of DRBD devices for disection etc. DRBD+LVM all your LVM structures are "magically" replicated on remote machine (including snapshots and other "wonders") I realize there are not many details above, but that's the point - it's a matter of preference in your environment and combinations are almost limitless (consider benefits of LVM-DRBD-LVM ;) ) -- Dmitry Makovey Web Systems Administrator Athabasca University (780) 675-6245 -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 189 bytes Desc: This is a digitally signed message part. URL: <http://lists.linbit.com/pipermail/drbd-user/attachments/20090313/a3fd60f3/attachment.pgp>