I've played with this quite a lot on a fairly similar machine to
yours - i.e. Dual pentium III 700s, 5x4.5Gb 10k Scsi disks Hardware
Raid0, 1 Gb. Ram, Windows 2000 Professional (SP3), Sql Server 2000
Desktop (SP2). As expected my elapse times are almost exactly twice as
long as yours.
I created my
[EMAIL PROTECTED] (Mark Davies) writes:
> The problem is, the self-join solution is extremely slow. I have a
> SQL Server 7.0 database with a clustered index on TheWord (sequential
> words) and a normal index on TheID.
Kindly do not pester Postgres mailing lists with SQL Server questions.
Mark,
I doubt very much you will ever get much faster results (without increasing
hardware) in a situation such as that. Your queries don't look selective
enough to effectively use the indexes. What is the query plan for each of
the individual selects and what does it look like as a whole? How
I have a database containing 100 million records, in which each record
contains (in sequence) all of the words in a 100 million word
collection of texts. There are two columns: TheID (offset value) and
TheWord (sequential words), e.g.:
TheID TheWord
-
1 I
2 saw
3 the
4 man
5
