On Fri, 2006-03-24 at 04:16, Jim C. Nasby wrote:
> On Thu, Mar 23, 2006 at 10:14:24AM +0100, Edoardo Serra wrote:
> > >Now, for the interesting test. Run the import on both machines, with
> > >the begin; commit; pairs around it. Halfway through the import, pull
> > >the power cord, and see which one comes back up. Don't do this to
> > >servers with data you like, only test machines, obviously. For an even
> > >more interesting test, do this with MySQL, Oracle, DB2, etc...
> >
> > I will surely run a test like this ;)
>
> If you do, I'd be *very* interested in the results. Pervasive would
> probably pay for a whitepaper about this, btw (see
> http://www.pervasivepostgres.com/postgresql/partners_in_publishing.asp).
Hehe. good luck with it.
At the last company I worked at I was the PostgreSQL DBA, and I could
not get one single Oracle, DB2, MySQL, MSSQL, Ingres, or other DBA to
agree to that kind of test.
6 months later, when all three power conditioners blew at once (amazing
what a 1/4" piece of wire can do, eh?) and we lost all power in our
hosting center, there was one, and only one, database server that came
back up without errors, and we know which one that was. No other
database there was up in less than 2 hours. So, I wandered the floor
watching the folks panic who were trying to bring their systems back
up.
And you know what? They still didn't want to test their systems for
recovery from a power loss situation.