NillaGoon Posted March 31, 2009 Report Share Posted March 31, 2009 I usually try to avoid operations that use the same physical disk as both source and destination, since this makes the disk spend most of its time moving the head around rather than reading and writing data. However, I'm moving into a configuration where physical disks are more closely intertwined because of RAID slicing and logical volume management. Separate logical volumes may in fact be partially stored on the same physical disk, so it's hard to keep backups and transfers "pure" in the usual way. I don't know if Retrospect uses any special code path to handle same-disk operations, but if it does, that path probably wouldn't be triggered under the new regime anyway since the connection to the physical disks is obscured. All Retrospect sees is network shares and iScsi volumes. And indeed, performance does seem to be notably worse. Is there any way to make Retrospect behave better in this situation by using a large memory buffer, say, several gigabytes? Reading a couple of gigabytes at once and then writing them out at once would be far more efficient than constantly trying to read and write simultaneously. Thanks, Garth Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.