Re: Any Good Way To Do Sync DB's?

On Fri, 12 Oct 2001, Joseph Koenig wrote:

> I have a project where a client has products stored in a large Progress
> DB on an NT server. The web server is a FreeBSD box though, and the
> client wants to try to avoid the $5,500 license for the Unlimited
> Connections via OpenLink software and would like to take advantage of
> the 'free' non-expiring 2 connection (concurrent) license. This wouldn't
> be a huge problem, but the DB can easily reach 1 million records. Is
> there any good way to pull this data out of Progess and get it into
> Postgres? This is way too large of a db to do a "SELECT * FROM table"
> and do an insert for each row. Any brilliant ideas? Thanks,

Yes. Throw away NT, and replace it with Linux or FreeBSD. They are free,
they are much more stable, and you will find that PostgreSQL on Linux is
likely to utterly blow away PostgreSQL on NT. IIRC, something to do with
the way NT handles forking and threads. Allegedly, this is particularly
noticeable in a heavy multi-user environment, where lots of connections
are spawned and killed.

Another thing you could try - Apache persistent DBI connections using
mod_perl. Set up Apache to use mod_perl and the persistent DBI database
connections through it. Then limit the number of Apache instances to 2
(i.e. up to 2 simultaneous connections). Then put an accelerating
squid proxy in front of the Apache web server, so that it can handle
static objects without wasting precious Apache back-ends.

All this seems like a hugely complicated solution when you could just
replace NT with Linux or FreeBSD and have a cheaper and MUCH faster
sollution, not to mention more stable.

Just my $0.02

Regards.

Gordan