From: | "Nathan Barnett" <nbarnett(at)cellularphones(dot)com> |
---|---|
To: | "'Robert B(dot) Easter'" <reaster(at)comptechnews(dot)com>, "'Nikhil G(dot) Daddikar'" <ngd(at)iname(dot)com>, <pgsql-general(at)postgresql(dot)org> |
Subject: | RE: Reliability Stability of PgSQL & it's JDBC driver |
Date: | 2000-12-14 14:56:21 |
Message-ID: | 71975481CD04D4118E57004033A2596E0DFB9E@ip205.82.136.216.in-addr.arpa |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
Robert,
I believe you are correct. An UPDATE is combination of INSERT and DELETE
and so would require vacuuming if you do it frequently.
---------------------------------------------
Nathan Barnett
-----Original Message-----
From: Robert B. Easter [mailto:reaster(at)comptechnews(dot)com]
Sent: Wednesday, December 13, 2000 1:19 PM
To: Nathan Barnett; 'Nikhil G. Daddikar'; pgsql-general(at)postgresql(dot)org
Subject: Re: [GENERAL] Reliability Stability of PgSQL & it's JDBC driver
On Wednesday 13 December 2000 10:05, Nathan Barnett wrote:
> Nikhil,
>
> Vacuuming the DB removes deleted entries from the database as they are not
> actually deleted when you use SQL to delete them, just marked as deleted.
> This operation clears up space and rearranges the data. If you don't
> perform many deletes, you may not need to run this. The only wrinkle is
> that you have to use the ANALYZE command to allow the database to generate
> statistics used in query planning. You should run this perdiodically
after
> a large number of inserts or deletes has taken place. This command must
be
> used in conjunction with VACUUM which slows down the database. There was
> talk of allowing the ANALYZE command to run independently of the VACUUM
> command, but I'm not sure when this will happen. I'm not completely clear
> on WAL functionality, but I believe that it automatically reuses the space
> taken up by deleted rows in the database. Someone on the list can
probably
> correct me if any of this is incorrect.
>
> ---------------------------------------------
> Nathan Barnett
Do UPDATEs also leave the old tuple behind to be cleaned out by VACUUM? If
so, then VACUUM is needed not only if you do many deletes, but also if you
do
lots of updates.
>
> -----Original Message-----
> From: Nikhil G. Daddikar [mailto:ngd(at)iname(dot)com]
> Sent: Wednesday, December 13, 2000 3:21 AM
> To: Nathan Barnett; pgsql-general(at)postgresql(dot)org
> Subject: Re: [GENERAL] Reliability Stability of PgSQL & it's JDBC driver
>
>
> Nathan,
>
> Thanks for your response. It is great to know that Postgres is working out
> fine. However, I don't understand what "Vaccum of the DB" means nor what
> the "WAL" functionality is. If you could briefly explain these, that would
> be great.
>
> Thanks again!
> -ngd.
>
> > The main problem that you have in a production system is the VACUUM of
> > the
>
> database.
>
> > This will in most cases severely inhibit performance. If your system
>
> isn't
>
> > 24x7, then you should be fine as you can schedule this for off times. I
> > believe that the need to vacuum the database will be removed in the next
> > version due to WAL functionality, but I'm sure someone else is better
> > qualified to explain that. I just wanted to vouch for the stability of
> > Postgres in a production environment.
--
-------- Robert B. Easter reaster(at)comptechnews(dot)com ---------
- CompTechNews Message Board http://www.comptechnews.com/ -
- CompTechServ Tech Services http://www.comptechserv.com/ -
---------- http://www.comptechnews.com/~reaster/ ------------
From | Date | Subject | |
---|---|---|---|
Next Message | Zolof | 2000-12-14 15:16:26 | Working under NT |
Previous Message | Tom Lane | 2000-12-14 14:53:56 | Re: space problem |