Re: Moving my business to PostgreSQL

> Sorry I can't answer too many of your questions, but one thing I am quite
> sure of is that Postgres does not offer replication/failover. I inferred
this

I did some more hunting tonight. It seems postgresql.com recently released a
beta of their eRServ, which looks to do exactly what I need in that reguard.

I know that it was high on the todo list, and then I heard something about
Great Bridge closing its doors so the replication work was loosing it's
sponsor. I also remember that someone out in academia had code that worked
with an older postgre version and was working on porting it to the current
codebase.

I was hoping that there was new information I hadn't come across :(.

> I really think that postgres is a great database; it certainly has many
> advanced features, and is also developing quite rapidly. However, it
sounds
> as though it doesn't have every capability you require.
Replication/failover
> seem important to what you're doing, and unless I'm mistaken, they don't
> exist in postgres. You may be able to make use of what work has been done
in
> that regard.

Indeed, it is important, simply because this business operates 24/7. I
simply have to have some way to run 2 servers at once so that everything can
fall over if one fails. There seem to be a couple middle ware things out
there designed to do connection pooling, but so far I can't find out for
sure if they support failover.

> I guess the question now is: Is there something out there that does meet
your
> requirements? Postgres may be your best bet if you can't afford the costs
> associated with oracle. Oracle may be best if the replication/failover
really
> is a requirement. Or maybe you can find a cheaper commercial database that
> offers more of what you need than postgres.

My options seem to be Postgre with a commercial replication product,
Interbase, building my own system with the sleepycat lib and some sort of
distributed que lib, or buying oracle. I don't like the idea of running MS
Sql on a remote host. I don't know oracle, and frankly, we'd spend more on
oracle liscensing than the business assests in their entirety cost us. We
anticipate around $1.2 mil in revenue this year, as the previous ownership
had been in cost freeze for a year, letting the business decline. We will
turn an ok profit accounting for the cost of the purchase, but what profit
we do achive needs to go back into growing the business, not software
liscenses.

So, really I guess I'm just in the position of praying Postgre gets
repliation soon. I'd like to think my c skills were hot enough I could grab
the tarball, add failover and get it into the cvs, but I'm not that arrogant
or naieve :P. Same argument applies to crufting something out of the Berkley
DB libs as well.

jason watkins