Re: full-text indexing

On Tue, 18 Apr 2000, Bruce Momjian wrote:

> Let me be specific.  The problem is that without cluster, your fragment
> rows are together in the index, but are all over the heap table, so you
> have to read in all those disk buffers, and that is slow.  With cluster,
> most of your matching fragments are on the same disk page, so one access
> gets them all.
>
> The nightly CLUSTER is a pain, but the only way I ever got it working
> quickly.

This is pretty painless compared to what I've had to do getting big
full-text systems like Excalibur to work.  I can deal with a nightly
cluster.

Brett W. McCoy
                                              http://www.chapelperilous.net
---------------------------------------------------------------------------
Twenty two thousand days.
Twenty two thousand days.
It's not a lot.
It's all you've got.
Twenty two thousand days.
        -- Moody Blues, "Twenty Two Thousand Days"