>>>Tom Lane said:
> Hm. Okay, so the number-of-rows estimate is not too far off. I concur
> with Hiroshi's comment: the reason the indexscan is so fast must be that
> the table is clustered (physical order largely agrees with index order).
> This would obviously hold if the records were entered in order by
> ipdate; is that true?
Yes. But... do you want me to cluster it by ipaddr for example and try it
again? I understand the clustering might help with sequential scans, but why
would it help with index scans?
Daniel