Re: Critical performance problems on large databases

Jean-Luc Lachance wrote:
> Bruce Momjian wrote:
> >
> > Oleg Bartunov wrote:
> > > The big issue with LIMIT,OFFSET is that it still use all rows
> > > for sorting. I already suggested to use partial sorting to avoid
> > > sorting all rows if one selected only first 20 row, for example.
> > > It's very important for Web applications because web users usually
> > > read first 1-2 pages. Our experimnets have shown 6 times performance
> > > win when using partial sorting.
> >
> > We do have this TODO item:
> >
> > * Allow ORDER BY ... LIMIT to select top values without sort or index
> > using a sequential scan for highest/lowest values
>
> It may not be wise to disregard the index if there is one for the ORDER
> BY.

Yes, this would be for cases where the index isn't useful, either
because ther isn't on or the join order makes it useless.

--
Bruce Momjian | http://candle.pha.pa.us
pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 853-3000
+ If your life is a hard drive, | 830 Blythe Avenue
+ Christ can be your backup. | Drexel Hill, Pennsylvania 19026