Hey there. 9.5.0.1536.
I had set up direct SAN access proxy (over FC 8 GiB) recently and pretty satisfied with it's performance. I surely saw the limitation of quick rollback restore over such type of proxy in the corresponding kb article, but I tried and it worked:
But it seems to work randomly o_O. I mean I`ve tried quick rollback 4 times and rollback was issued only twice. Don't know what should I think about it.
Also I would like to ask about poor restore performance over SAN. The only legitimate way to increase performance is to make eager zeroed disks for vms, isn't it?
I`ve noticed that I can't do quick rollback even over NBD proxy actually. I checked quick rollback flag and full vm restoration was issued (with full disk restoration). I`m scared now.
update: found post257035.html but still don't get why I can't use quick rollback few times in a row.
Ok - sorry. From what i understand CBT is not tracking the changes done by the restore itself. Because of that a second restore can not know what the changestate of the blocks on the production datastore is and thus is doing a full restore.
Yeah, I got it but I cant imagine how it works, I mean I do not understand the explanation that Andreas Neufert described in this post post257035.html and I can't find any intuitive scheme.
