Hi Roy,
I have been reading the archives and it appears the MSSQL backer is busted, can
anyone confirm this? I am stuck with a couple legacy 2000 boxes but building out
in addition with Postgresql.
I think that the only thing that the MSSQLBackuper requires is a shared
storage between replicas to store the backup.
Otherwise you can still use the ScriptBackuper and invoke your own
backup/restore tools.
Additionally I have a conceptual question which I think I know the answer to but
can't hurt to ask. I am assuming that the backends themselves have no idea if
the databases are the same between them unless it's data written by the
controller
is that correct? What happens if some process like a stored procedure changes
data
on one machine, will a replication mechanism be required to all the backends
consistent?
If you invoke the stored procedure though Sequoia, the call will be
propagated in total order. You must just ensure that the stored
procedure execution is deterministic so that each backend writes the
same data.
In general, if you bypass Sequoia to execute requests that update the
database (includes stored procedures, user defined functions, sequences,
environment variables, triggers...), the cluster consistency is not
ensured anymore.
Hope this helps,
Emmanuel
--
Emmanuel Cecchet - Research scientist
EPFL - LABOS/DSLAB - IN.N 317
Phone: +41-21-693-7558
_______________________________________________
Sequoia mailing list
[email protected]
https://forge.continuent.org/mailman/listinfo/sequoia
