On Tue, 27 Jul 1999, Zeugswetter Andreas IZ5 wrote:
> Date: Tue, 27 Jul 1999 16:57:06 +0200
> From: Zeugswetter Andreas IZ5 <Andreas.Zeugswetter@telecom.at>
> To: 'Oleg Bartunov' <oleg@sai.msu.su>
> Cc: "'hackers@postgresql.org'" <hackers@postgresql.org>
> Subject: Re: [HACKERS] UPDATE performance degradation (6.5.1)
>
>
> > This is for table with one row after a lot of updates.
> > Too much. vacuum analyze this table was a good medicine !
> > Is this a design problem ?
> >
> In PostgreSQL an update always adds a new row to the table.
> The old rows get eliminated by vacuum that is the whole business of vacuum.
> There has been some discussion for implementing row reuse,
> but that is a major task.
Ok, I understand now the size of the table. What's about index file ?
Why it's so big. Look. just did delete from hits and vacuum analyze.
om:/usr/local/pgsql/data/base/discovery$ l hits*
-rw------- 1 postgres users 0 Jul 27 19:14 hits
-rw------- 1 postgres users 2015232 Jul 27 19:14 hits_pkey
after 6500 updates:
om:/usr/local/pgsql/data/base/discovery$ l hits*
-rw------- 1 postgres users 344064 Jul 27 19:23 hits
-rw------- 1 postgres users 2097152 Jul 27 19:23 hits_pkey
and it took a lot of time. Also I populate table hits by 10,000 rows
and run the same test. It was incredibly slow.
It seems index file doesn't affected by vacuum analyze !
Could we consider this as a bug ?
Regards,
Oleg
>
> Andreas
>
_____________________________________________________________
Oleg Bartunov, sci.researcher, hostmaster of AstroNet,
Sternberg Astronomical Institute, Moscow University (Russia)
Internet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/
phone: +007(095)939-16-83, +007(095)939-23-83