|Subject||Re: Some enhancements to NBackup to allow backups more often|
> Yes, I had a discussion with Jim some time ago about this but theI cannot say anything about SMB, should work the same way. On the
> thing remained in the fog... And also isn't only about NFS is about
> also Window's \\<Server>\<Share> way. (We have the servers on the
> Win platform). This will be the greatest way to keep the db safe on
> another system in a case of a very serious of hardware failure
> (let's say we have a special multiprocessor motherboard).
other hand the "Shadow-on-NFS" is already used by Alex Peshkov, but I
don't know about anybody who tried that with SMB. So the reliability
of such thing is not yet known.
> Can you give a timeline for this?No. We have discussed the required pieces mainly related to how to
detect the node failure and detach the shadow, so it is not corrupted
by the failed node that still is able to write into the shadow
("fail-stop" failures do not require such care). The thing looks
doable. So, we need a taker and/or a sponsor for the feature.
And one more idea for you - clustered mode a-la Oracle RAC. Firebird
is theoretically extendable to support the external DLM (Jim says it
worked in the old days, but many things have changed since then). Now
the OpenDLM (and its continuation in Red Hat Linux) seems to be
compatible with the interfaces used in Firebird. So one needs to teach
Firebird to use OpenDLM and you can then use NAS to store database
there and run your Firebird instances on Windows box (the DLM itself
would require RH Linux though). Also seems doable. Again we need a
taker and a sponsor.