Re: Speed while runnning large transactions.

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: "Kevin Grittner" <Kevin(dot)Grittner(at)wicourts(dot)gov>
Cc: "Greg Smith" <gsmith(at)gregsmith(dot)com>, jesper(at)krogh(dot)cc, pgsql-performance(at)postgresql(dot)org
Subject: Re: Speed while runnning large transactions.
Date: 2009-10-02 22:01:05
Message-ID: 15124.1254520865@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-performance

"Kevin Grittner" <Kevin(dot)Grittner(at)wicourts(dot)gov> writes:
> Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> As of 8.4, the typical case is that an open transaction blocks
>> deletion of rows that were deleted since the transaction's current
>> *statement* started.

[ BTW, of course that should have read "blocks removal of" ... ]

> Surely the original version of a row updated or deleted by the
> long-running transaction must be left until the long-running
> transaction completes; otherwise, how does ROLLBACK work?

Right. What I was talking about was the impact of a long-running
transaction on the removal of rows outdated by *other* transactions.
The people who hollered loudest about this seemed to often have
long-running read-only transactions in parallel with lots of short
read-write transactions. That's the pattern that 8.4 can help with
anyway.

regards, tom lane

In response to

Responses

Browse pgsql-performance by date

  From Date Subject
Next Message Justin Pryzby 2009-10-03 01:48:08 dump time increase by 1h with new kernel
Previous Message Dimitri Fontaine 2009-10-02 21:58:15 Re: Best suiting OS