From: | Marti Raudsepp <marti(at)juffo(dot)org> |
---|---|
To: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
Cc: | Robert Haas <robertmhaas(at)gmail(dot)com>, maxim(dot)boguk(at)gmail(dot)com, pgsql-bugs(at)postgresql(dot)org |
Subject: | Re: BUG #6513: explain (analyze, buffers) and toasted fields |
Date: | 2012-03-12 19:41:48 |
Message-ID: | CABRT9RCAUObsWMG9KBhmeaKgvjC22QA+OA8aNOv-kXRLHzO13g@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-bugs |
On Mon, Mar 12, 2012 at 17:23, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> EXPLAIN ANALYZE *necessarily* does less work than the real query,
> because it doesn't transmit the results to the client (which is going
> to be a dominant cost in a lot of situations). I'm not sure whether
> running the I/O functions would provide a useful improvement in
> verisimilitude or not
Granted, but I can see how one would expect the detoasting to be
accounted for in EXPLAIN, since it's something that's usually done
*inside* the database (as opposed to network overhead). That would
give a more accurate picture of where the time is being spent.
Now, another question is whether to use text or binary output
functions. And where to account the time, since this happens outside
of the plan tree.
At the very least, I think this gotcha should be documented on the EXPLAIN page.
Regards,
Marti
From | Date | Subject | |
---|---|---|---|
Next Message | Sergey Burladyan | 2012-03-12 21:10:42 | Re: BUG #6480: NLS text width problem |
Previous Message | samuel.gilbert | 2012-03-12 19:34:20 | BUG #6527: Error in COPY documentation syntax |