Informix is a SQL engine targeted at the DataWarehouse market. It was designed to do big jobs from the word go.
I've only tinkered with the "big iron" implementation, I know zilch about the Unix product and I don't know if Informix even runs on the NT platform.
The Informix client tools can be charitably described as "interesting". While Informix included a lot of nice API features (especially for Red Brick), it wasn't long on creature comforts for the folks that had to use it directly.
The primary intent was to load data via the API, coming from transaction oriented systems of one sort or another or via an ETL tool. I don't have any experience with Informix loaders provided with the engine.
I liked Informix. But I was on Standard Engine 4.1 or some such. Vintage '92, or something.
Informix missed out on the Linux boat in a big way, and paid for it. The database product was bought by IBM, and the rest of the company renamed itself Ascential Software. Not exactly sure what they do.
I'm Informix and DB2 certified - but only because clients like paper on the wall.
Q. Well...what's it like?
A. Runs on multiple OS platforms - windows, AIX, Unix, Linux, Sun, etc.. Configurable from small to very large. Provides a lot of bang for the buck in terms of throughput rate on given set of iron. Handles concurrent user connections in the thousands.
Q. Somewhere between sql server and oracle?
A. Like between a rock and a hard place? Informix arch is very centralized. Memory (including buffers), logs, backups/restores all at the instance level.
Q. Or is it a totally different animal?
A. Since Informix has always been multi-platform some of the stuff you windows jockeys take for granted are not there. Nor would the be there in DB2 or Oracle. Informix developers always concentrated on two things, 1) works the same across platforms and 2) performance.
Q. Just found out it's an IBM family product so I wonder if it's like DB2 at all
A. No. That's one of the reasons IBM never saw a rush to have Informix sites convert to DB2. While Informix is currently being enhanced the IBM strategy is to dovetail the two products using a "bets of breed" approach.
Q. What's the client tools like?
A. The client system requires an interface loaded (setnet32) but then can use any standard odbc product.
Q. An immediate question would be how do the load data?
A. Many options here based upon what the data looks like. Standard delimited files could use something as simple as dbaccess with a "load from ... insert into tablename" or several high perrformance loaders. Peer to peer would require setnet32 on your sql box.