Re: Point in time recovery?

On Mon, Mar 24, 2003 at 04:06:46PM -0500, Robert Treat wrote:

> data corruption in the first place. Problems like power outages and
> server crashes (at either the os/kernel/application level) are almost
> certainly not going to cause data loss. I could see possible issues with

This is certainly true.  We have never, not once, had a problem with
PostgreSQL losing production data.  It is hard to break even if you
try (although I've managed a couple times in tests, with enough
ingenuity.  It took more ingenuity than other things I've had to
break, mind you).  But insurance is insurance, and telling managers,
clients, or regulatory bodies, "Don't worry, PostgreSQL never
breaks," is not going to fly.

One PITR-provided advantage that you cannot get with the replication
approach, by the way, is this one: "Junior DBA issued 'DELETE FROM
million_dollar_table;' at 05:00 on Saturday.  I want to go back to
04:59."  Your replicated database will have the same problem as the
main one.

A

--
----
Andrew Sullivan                         204-4141 Yonge Street
Liberty RMS                           Toronto, Ontario Canada
<andrew@libertyrms.info>                              M2P 2A8
                                         +1 416 646 3304 x110