I've been following this thread with interest since it started, and it
really seems that there is just too much data in that single table. When
it comes down to it, making smaller separate tables seems to be the way
to go. Querying will be a little harder, but much faster.
Warmest regards,
Eri
On Sat, 03 Apr 2004 22:39:31 -0800, "Mooney, Ryan" <[EMAIL PROTECTED]>
wrote:
>Ok, so I ran a vacuum analyse. It took ~1.7 days to finish.
Just to make it clear: VACUUM and ANALYSE are two different commands.
VACUUM is for cleaning up. It has to visit every tuple in every page,
and if there
[time to move this to -hackers]
On Fri, 02 Apr 2004 11:16:21 -0500, Tom Lane <[EMAIL PROTECTED]> wrote:
>Manfred Koizar <[EMAIL PROTECTED]> writes:
>> The first step, however, (acquire_sample_rows() in analyze.c) has to
>> read more rows than finally end up in the sample. It visits less than
>> O
Manfred Koizar <[EMAIL PROTECTED]> writes:
> The first step, however, (acquire_sample_rows() in analyze.c) has to
> read more rows than finally end up in the sample. It visits less than
> O(nblocks) pages but certainly more than O(1).
> A vague feeling tries to tell me that the number of page rea
On Thu, 01 Apr 2004 12:22:58 +0200, I wrote:
>BTW, ANALYSE is basically a constant time operation.
On closer inspection, this is not the whole truth. ANALY[SZ]E is a two
stage process: First it collects a sample of rows, then these rows are
examined to produce various statistics.
The cost of th
On Tue, 30 Mar 2004 17:48:14 -0800, "Mooney, Ryan" <[EMAIL PROTECTED]>
wrote:
>I have a single table that just went over 234GB in size with about 290M+
>rows.
That would mean ~ 800 bytes/row which, given your schema, is hard to
believe unless there are lots of dead tuples lying around.
>queries u
The issue here might be just organizing the data differently. Or getting
an Opteron server with 16GB RAM :-) Based on the strength of the
developers recommendations in this newsgroup, we recently upgraded to a
dual Opteron 2GHZ with 16GB Ram and 15K hard drives. We set
shared_buffers to 40,000
Hello,
I have a single table that just went over 234GB in size with about 290M+
rows. I think that I'm starting to approach some limits since things
have gotten quite a bit slower over the last couple days. The table is
really simple and I'm mostly doing simple data mining queries like the
quer
> Any suggestion on how to prepare for the next upgrade would be
> appreciated.
I think it has to be said that if you want decent performance on excessively
large (50GB+) databases, you're going to need excessively good hardware to
operate it on. Buy a 3ware IDE RAID controller (www.hypermicro.
Philip Crotwell wrote:
> Hi
>
> I have a very large database of seismic data. It is about 27 Gb now, and
> growing at about the rate of 1 Gb every 3-4 days. I am running
Out of curiosity, how long does it take you to vacuum that?
--
Joseph Shraibman
[EMAIL PROTECTED]
Increase signal to nois
Hi
I have a very large database of seismic data. It is about 27 Gb now, and
growing at about the rate of 1 Gb every 3-4 days. I am running
postgres 7.1.2. I might possibly try to upgrade to 7.2 when it comes out,
but I don't know if it will be possible for me to do 7.3 due to
the pg_dump/pg_rest
11 matches
Mail list logo