Re: Informix limitations, should we be using Oracle?

Simon M. wrote:
> Well, what can I say??? Didn't mean to start WWIII guys! First of,

Well, then, why did you post so much "competitive analysis"?

> thanks to Daniel for posting the only useful response. To the others,

What? "If there is no rebuttal, then Oracle is the way to go?" Well, I rebutted, so that means if my response wasn't useful, you don't want to use XPS.

> I can assure you that I don't work for Oracle or any marketing company
> or whatever. I've picked up a large DW project that has been mangled
> in the past and I'm trying to find out some of the root causes. Of

Well, the root causes are that someone made some very bad decisions and is now trying to find a scapegoat, and XPS looks like a good scapegoat. XPS on a 4-way single node is never going to reach its full potential. The software or hardware in isolation are never the problem. If you were implementing *any* RDBMS on that platform with those sort of data volumes, I'd be "curious" as to how the decision was reached. But I'm not saying the tin is the problem. I'm saying the tin is wrong for XPS; the combination is wrong.

> course I'm going to IBM as well to get their feedback but as I've had
> some good help from this group before, I thought that maybe someone
> would have some relevant experience or could help to set me and the
> team here straight.

And? Did we?

> We've tried to get some experienced resource in to help us out but out
> here in Australia there are no appropriate skills. None. We've tried.

Ask IBM. Tell them it's a CritSit.

> It's a last gasp for us... I'd love to get the thing working as it's
> intended but at most we've been able to read the manual, make
> observations on performance and draw our own conclusions - which from
> the feedback, seem to be fundamentally wrong. No-one has pointed out
> why, but not to worry, we'll keep trying.

I did. And you really need to upgrade, especially given some of your requirements.

> Oh, completely agree about the 'why are you running XPS on one node?'
> comments. That is what we inherited and we don't understand either...
> if someone could give me the $$$ then I'd go out and buy more servers
> etc. but that ain't going to happen in a hurry.

But you (not you personally, of course!) have got the money to buy Oracle and get everyone trained and migrate the data and....? Hmmm...

> Our feeling at the moment is that we need a db for which there are DBA
> skills available on the open market in our country. At the moment, we
> can't see what's going on in XPS and we can't see that changing...

The skills transfer process doesn't take that long. Just get in a consultant. IBM's white paper on Informix covers XPS quite clearly, I believe.

> Sorry to waste anyone's time and feel free to ignore the thread!

Feel free to ignore my useless posts.

> (Oh, and yes, of course it is a star schema...)

Look at RedBrick then. It will run far better on a single node than XPS. But you have to be *really* religious about the star schema. And RBW has it's own quirks, too. Don't try to implement it without some kind of knowledge transfer either. Received on Thu Oct 17 2002 - 09:13:13 CEST