| From: | Robert Haas <robertmhaas(at)gmail(dot)com> |
|---|---|
| To: | Craig Ringer <craig(at)postnewspapers(dot)com(dot)au> |
| Cc: | Thom Brown <thombrown(at)gmail(dot)com>, "Schmitz, David" <david(dot)schmitz(at)harman(dot)com>, Andres Freund <andres(at)anarazel(dot)de>, pgsql-performance(at)postgresql(dot)org |
| Subject: | Re: performance penalty between Postgresql 8.3.8 and 8.4.1 |
| Date: | 2009-12-08 15:02:05 |
| Message-ID: | 603c8f070912080702j1a1df335v6932660b0afe05f2@mail.gmail.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-performance |
On Tue, Dec 8, 2009 at 7:12 AM, Craig Ringer
<craig(at)postnewspapers(dot)com(dot)au> wrote:
> On 8/12/2009 6:11 PM, Thom Brown wrote:
>
>> Your output shows that the xdf_admin_hierarchy tables between versions
>> are drastically different. 8.3.8 only contains 1 row, whereas 8.4.1
>> contains 84211 rows.
>
> That's just because one of them is doing a nested loop where it looks up a
> single row from xdf_admin_hierarchy via its primary key on each iteration.
> The other plan is doing a hash join on a sequential scan over
> xdf_admin_hierarchy so it reports all the rows at once.
I've been meaning to write a patch to show the places after the
decimal point in that case. Rounding off to an integer is horribly
misleading and obscures what is really going on. Although in this
case maybe it would come out 1.000 anyway.
...Robert
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Kevin Grittner | 2009-12-08 15:05:04 | Re: Checkpoint spikes |
| Previous Message | Matthew Wakeling | 2009-12-08 14:48:33 | Re: Optimizing Bitmap Heap Scan. |