From: | Graham Davis <gdavis(at)refractions(dot)net> |
---|---|
To: | pgsql-performance(at)postgresql(dot)org |
Subject: | Re: BUG #2658: Query not using index |
Date: | 2006-10-03 18:20:49 |
Message-ID: | 4522AA01.4020709@refractions.net |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-bugs pgsql-performance |
Hi,
Adding DESC to both columns in the SORT BY did not make the query use
the multikey index. So both
SELECT DISTINCT ON (assetid) assetid, ts
FROM asset_positions
ORDER BY assetid, ts DESC;
and
SELECT DISTINCT ON (assetid) assetid, ts
FROM asset_positions
ORDER BY assetid DESC, ts DESC;
use the same query plans and both do sequential scans without using either the (assetid, ts) or (ts) indexes. Any other ideas on how to make this query use an index? Thanks,
--
Graham Davis
Refractions Research Inc.
gdavis(at)refractions(dot)net
>On Wed, Sep 27, 2006 at 20:56:32 +0000,
> Graham Davis <gdavis(at)refractions(dot)net> wrote:
>
>
>>SELECT assetid, max(ts) AS ts
>>FROM asset_positions
>>GROUP BY assetid;
>>
>>I have an index on (ts), another index on (assetid) and a multikey index on
>>(assetid, ts). I know the assetid index is pointless since the multikey one
>>takes its place, but I put it there while testing just to make sure. The
>>ANALYZE EXPLAIN for this query is:
>>
>> QUERY PLAN
>>----------------------------------------------------------------------------
>>-------------------------------------------------------------
>> HashAggregate (cost=125423.96..125424.21 rows=20 width=12) (actual
>>time=39693.995..39694.036 rows=20 loops=1)
>> -> Seq Scan on asset_positions (cost=0.00..116654.64 rows=1753864
>>width=12) (actual time=20002.362..34724.896 rows=1738693 loops=1)
>> Total runtime: 39694.245 ms
>>(3 rows)
>>
>>You can see it is doing a sequential scan on the table when it should be
>>using the (assetid, ts) index, or at the very least the (ts) index. This
>>query takes about 40 seconds to complete with a table of 1.7 million rows.
>>I tested running the query without the group by as follows:
>>
>>
>
>
>
>>SELECT DISTINCT ON (assetid) assetid, ts
>>FROM asset_positions
>>ORDER BY assetid, ts DESC;
>>
>>
>
>This is almost what you want to do to get an alternative plan. But you
>need to ORDER BY assetid DESC, ts DESC to make use of the multicolumn
>index. If you really need the other output order, reverse it in your
>application or use the above as a subselect in another query that orders
>by assetid ASC.
>
>
From | Date | Subject | |
---|---|---|---|
Next Message | ragetron99 | 2006-10-03 18:39:00 | BUG #2672: stored procedure argument and return type length validation |
Previous Message | Simon Riggs | 2006-10-03 18:13:06 | Re: BUG #2671: incorrect return value by RULE |
From | Date | Subject | |
---|---|---|---|
Next Message | Chris Browne | 2006-10-03 19:02:32 | Re: BUG #2658: Query not using index |
Previous Message | Carlo Stonebanks | 2006-10-03 18:07:22 | Re: Performance Optimization for Dummies 2 - the SQL |