Re: SE and big databases or RAID

In article <4gatrn$4qv@cssun.mathcs.emory.edu> mac@antares-dev.com (Miguel A. Caldera) writes: > >I know this is will generate remarks like "Use Online, stupid!", but... Not necessary, in my experience. > >I have a customer running SE for the obvious cost savings over Online. >We are projecting a database size in the 2 to 4 million rows size for >one of the big tables. The records are not large, 15 rows for a total of >126 bytes. The database will probably take over a year to get >there, but it is expected to get that big. Has anybody ever run SE that >large? I expect to see some performance degradation, but the customer >is prepared to throw more money at the hardware before he will hire >someone to maintain Online. The customer already has 250,000 records >and is very pleased with current performance. I have one database in SE Ver 5.02 on a SPARC 5 that has 750K rows in the primary table (patients) and 2.8 million rows in the largest table, total size approx 800MB at the moment. Grows at the rate of 100K patients, 4-500K test results per year. It's used 5 days per week, every week and has never missed a beat. We love it. OnLine might be better, but no-one there wants the extra aggravation. My wife runs the lab and with SE, her routine maint. is putting in a tape every night. Recently, I added a new column to all 3 main tables (patient, sample & result) hitting over 4 million rows. This completed in about 4 hours. Used up one hell of a lot of my 2Gb scratch disk in the process, though :-). Peter Wiley