Sequential scans on a one row table ?

Hi all Just had a look at my "onstat -p" and it was showing a terribly large number of seqscans (~100K), so I went in and had a look at the offending table. Turns out the table with the most seqscans against it has only got *one row* in it. There were some other tables having seqscans done on them, ranging in size from about 10 to 10K rows. My setup is: IDS 7.24.UC6 on Solaris 2.6 on a Sun E450 (3 CPU) with 1GB RAM. The database is on cooked UFS disks. I run update stats high nightly (probably overkill but hey I've got some spare CPU cycles so why not:) My questions are: 1. Is it just a recording error that a one row table can contribute to the seqscans count ? Would I be correct in discounting any seqscans on a one row table when I'm looking at things like sysptprof in SMI ? 2. How many rows should a table be before sequential scans on that table hurt performance ? eg if a 100 row table was being seqscanned, would you be inclined to put an index on that table as a matter of course ? Thanks! Andrew Reardon Boeing Australia Tel: +61 7 3306 3346 Mobile: 0419 745 831