Hi All,
Agreed, snapshots are not backups - but in this context they are not intended
as such - simply as a consistency mechanism by which a backup can be made.
From further reading about Hyper-V on Windows 2012R2, this renames the
snapshots as checkpoints and furthermore these checkpoints provide support for
Linux VMs without the need for VSS support on the O/S or applications. So - I
think(!) - this makes the VM backup crash-consistent; including the Oracle
database running on it. So not quite as complete a solution as backing up a
Windows VM, where you can have application consistency, but possibly good
enough for our purposes provided we're happy to endure crash recovery and loss
of non committed data (which we are).
Obviously, as pointed out, this requires thorough testing. But I'm still
really looking for confirmation on the concepts - I can see scenarios where it
might work by fluke during on a test - but not in reality if it's not meant to
work that way.
Thanks!Charlotte
On Wednesday, August 29, 2018 10:17 AM, Stefan Koehler <contact@xxxxxxxx>
wrote:
Hello Charlotte,
snapshots are NO backup! You don't have any physical or logical corruption
checks by doing snapshots on storage or VM level and so you are screwed if you
discover some of these corruptions later on. Snapshots are fine in case of
upgrade scenarios or so if you just want to revert very quickly to a specific
point in time but are not a proper solution for continuous backup.
In addition you may experience VM freezing problems while doing snapshot
backups if your database load increases, e.g. like this one I have troubleshot
with Veeam and VMware (https://www.veeam.com/kb1681) at a client site.
NetBackup (or BackupExec) also got a MML for RMAN.
Best Regards
Stefan Koehler
Independent Oracle performance consultant and researcher
Website: http://www.soocs.de
Twitter: @OracleSK
Charlotte Hammond <dmarc-noreply@xxxxxxxxxxxxx> hat am 28. August 2018 um--
20:22 geschrieben:
Hello All,
I've been handed an Oracle environment using technologies that I'm unfamiliar
with and am looking for guidance on first steps. These are Oracle 11g/12c
databases running on Linux VMs in Hyper-V.
The Hyper-V VMs are backed up using NetBackup which appears to use a
temporary Snapshot (Checkpoint) of the entire VM followed by a backup of this
snapshot.
The databases are used only for testing and have a lax RPO - simply go back
to the last snapshot - no roll-forward required.
The question is what do I need to do? Is simply allowing NetBackup to backup
Hyper-V snapshots adequate for protecting the VM contents including (most
importantly) the Oracle database? If so is this entirely transparent to the
VM / database? Is there a requirement to put the database into backup mode
during the snapshoting (and if so, how)? Does the snapshot capture memory or
just disk content - i.e. will this require crash recovery on restore?
Obviously testing will be required but I'm struggling to find appropriate
information on concepts just to get started which suggests I'm asking the
wrong questions, so any help will be very much appreciated!
Thank You!
Charlotte