Re: Looking for recommendation of using informix for a large amount of data

> We are about to begin evaluating informix (currently using oracle). Our > oracle database is about 50 gig in size and we are having performance > issues with queries. We have a denormalized schema to avoid joining > tables. We have 3 main tables (customers - 12 million records, orders - > 27 million records, items - 80 million records). Raw data is about 15 > gig. Any advice or similar experience responses would be appreciated. I've unfortunately never worked with that many records (as far as I'm aware :) However, I have scaled up an artificial benchmark to 1 million records, purely as a test. I found that my eye could not detect the difference in speed between an indexed search of 10000 records and 1 million (there was a fractional, but inconsistent difference reported by timex). As long as your hardware is "big" enough and you have the correct indicies, selecting 1 or few records should pose no problems. If, however, you have to trawl the entire table, I've found it far more machine dependent. You *can* tune the engine, but unfortunately I didn't have enough time to assess the impact. Judging by your volumes, I would advise considering OnLine 7 on a proper SMP machine with >=1GB RAM and (personal preference) no RAID. Vague, I know, but HIH. -- Ciao, Billy +------------------------------------------------------------------------------+ | Billy Wheeler Junior Programmer / Director The West Solutions Group | +------------------------------------------------------------------------------+ | The Informix FAQ is at: http://www.iiug.org/techinfo/faq/faq_top.html | +------------------------------------------------------------------------------+ | .... oh, and Kerry, I see it's _still_ not at rtfm.mit.edu... :-) | +------------------------------------------------------------------------------+