MIME-Version: 1.0
References: <CA+7GS+Ff3PWbeJNbi7e6Nqfgh4HOTzyNoyGsBwFeivkUKDUwBA@mail.gmail.com>
In-Reply-To: <CA+7GS+Ff3PWbeJNbi7e6Nqfgh4HOTzyNoyGsBwFeivkUKDUwBA@mail.gmail.com>
From: =?UTF-8?Q?Torsten_F=C3=B6rtsch?= <tfoertsch123@gmail.com>
Date: Fri, 30 Aug 2024 16:42:43 +0200
Message-ID: <CAKkG4_ncUgaNGRCNpwQiL1tWkkWpByiLYZD2hU8r2WGzDWM=dQ@mail.gmail.com>
Subject: Re: optimizing a join against a windowed function
To: James Brown <james@instrumentl.com>
Cc: pgsql-general@lists.postgresql.org, Gabriel Sinkin <gabe@instrumentl.com>
Content-Type: multipart/alternative; boundary="00000000000050a7bd0620e79b78"
Archived-At: <https://www.postgresql.org/message-id/CAKkG4_ncUgaNGRCNpwQiL1tWkkWpByiLYZD2hU8r2WGzDWM%3DdQ%40mail.gmail.com>
Precedence: bulk

--00000000000050a7bd0620e79b78
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

Try perhaps something along these lines:

```
SELECT t.id, i.insight_id
  FROM taxpayers AS t
  JOIN LATERAL (
           SELECT x.id AS insight_id
             FROM insights AS x
            WHERE x.taxpayer_id =3D t.id
              AND x.year IS NOT NULL
            ORDER BY year DESC
            LIMIT 1
       ) AS i ON true
 WHERE t.id IN (?, ?)
```

If you don't have millions of ? in that IN clause, then that might be
faster.

On Fri, Aug 30, 2024 at 1:36=E2=80=AFPM James Brown <james@instrumentl.com>=
 wrote:

> Hello:
>
> I'm attempting to figure out whether an optimizer behavior I'm seeing is =
a
> PostgreSQL bug or expected behavior. The scenario:
>
> I have two tables: one named taxpayers which has a goodish number of
> columns an an integer PK id, and one named insights, which has
> a taxpayer_id foreign key to taxpayers, a year, and (again) a lot of othe=
r
> columns. There's an index on insights (taxpayer_id, year DESC). I'm
> executing the following SQL:
>
> ```
> SELECT taxpayers.id, insight_id
>    FROM taxpayers
>    JOIN (
>        WITH ordered_insights AS (
>            SELECT taxpayer_id, id, RANK() OVER (PARTITION BY taxpayer_id
> ORDER BY year DESC) AS rank
>            FROM insights
>            WHERE year IS NOT NULL
>        )
>        SELECT taxpayer_id, id AS insight_id
>        FROM ordered_insights
>        WHERE rank =3D 1
>    ) latest_insights ON latest_insights.taxpayer_id =3D taxpayers.id
> WHERE taxpayers.id IN (?, ?)
> ```
>
> (this is simplified example; the real code has the subselect in a view so
> that it can execute this kind of join from an ORM; it also joins quite a
> few tables downstream after this)
>
> If there's only a single value in the IN clause, the EXPLAIN plan looks
> great:
>
> Nested Loop  (cost=3D0.86..53.30 rows=3D1 width=3D16)
>   ->  Index Only Scan using taxpayers_pkey on taxpayers  (cost=3D0.43..8.=
45
> rows=3D1 width=3D8)
>         Index Cond: (id =3D 650974)
>   ->  Subquery Scan on ordered_insights  (cost=3D0.43..44.83 rows=3D1 wid=
th=3D16)
>         Filter: (ordered_insights.rank =3D 1)
>         ->  WindowAgg  (cost=3D0.43..44.71 rows=3D10 width=3D28)
>               Run Condition: (rank() OVER (?) <=3D 1)
>               ->  Index Scan using index_insights_on_taxpayer_id_year_des=
c
> on insights  (cost=3D0.43..44.53 rows=3D10 width=3D20)
>                     Index Cond: ((taxpayer_id =3D 650974) AND (year IS NO=
T
> NULL))
> (9 rows)
>
> However, if there are multiple rows in the IN clause, the optimizer
> decides to execute the subselect against the entire giant table, and it i=
s
> not great:
>
> Hash Join  (cost=3D2611586.97..2800201.15 rows=3D1 width=3D16)
>   Hash Cond: (ordered_insights.taxpayer_id =3D taxpayers.id)
>   ->  Subquery Scan on ordered_insights  (cost=3D2611570.10..2799818.65
> rows=3D28961 width=3D16)
>         Filter: (ordered_insights.rank =3D 1)
>         ->  WindowAgg  (cost=3D2611570.10..2727415.36 rows=3D5792263 widt=
h=3D28)
>               Run Condition: (rank() OVER (?) <=3D 1)
>               ->  Sort  (cost=3D2611570.10..2626050.76 rows=3D5792263 wid=
th=3D20)
>                     Sort Key: insights.taxpayer_id, insights.year DESC
>                     ->  Seq Scan on insights  (cost=3D0.00..1723354.01
> rows=3D5792263 width=3D20)
>                           Filter: (year IS NOT NULL)
>   ->  Hash  (cost=3D16.85..16.85 rows=3D2 width=3D8)
>         ->  Index Only Scan using taxpayers_pkey on taxpayers
>  (cost=3D0.43..16.85 rows=3D2 width=3D8)
>               Index Cond: (id =3D ANY ('{650974,243848}'::bigint[]))
>
> If I add in a second repetitive WHERE clause, it goes back to being happy=
,
> but that feels a bit like a hack:
>
> # EXPLAIN SELECT taxpayers.id, insight_id
>    FROM taxpayers
>    JOIN (
>        WITH ordered_insights AS (
>            SELECT taxpayer_id, id, RANK() OVER (PARTITION BY taxpayer_id
> ORDER BY year DESC) AS rank
>            FROM insights
>            WHERE year IS NOT NULL
>        )
>        SELECT taxpayer_id, id AS insight_id
>        FROM ordered_insights
>        WHERE rank =3D 1
>    ) latest_insights ON latest_insights.taxpayer_id =3D taxpayers.id
> WHERE taxpayers.id IN (650974, 243848) AND latest_insights.taxpayer_id IN
> (650974, 243848);
>                                                         QUERY PLAN
>
>
> -------------------------------------------------------------------------=
----------------------------------------------------
> Nested Loop  (cost=3D0.86..110.57 rows=3D1 width=3D16)
>   Join Filter: (taxpayers.id =3D ordered_insights.taxpayer_id)
>   ->  Subquery Scan on ordered_insights  (cost=3D0.43..93.69 rows=3D1 wid=
th=3D16)
>         Filter: (ordered_insights.rank =3D 1)
>         ->  WindowAgg  (cost=3D0.43..93.42 rows=3D21 width=3D28)
>               Run Condition: (rank() OVER (?) <=3D 1)
>               ->  Index Scan using index_insights_on_taxpayer_id_year_des=
c
> on insights  (cost=3D0.43..93.06 rows=3D21 width=3D20)
>                     Index Cond: ((taxpayer_id =3D ANY
> ('{650974,243848}'::bigint[])) AND (year IS NOT NULL))
>   ->  Index Only Scan using taxpayers_pkey on taxpayers  (cost=3D0.43..16=
.85
> rows=3D2 width=3D8)
>         Index Cond: (id =3D ANY ('{650974,243848}'::bigint[]))
>
> This feels like a bug to me, but maybe I'm missing something obvious. I
> don't really get why the optimizer wouldn't be able to infer the second
> condition given that I'm doing a join on a non-nullable integer column (s=
o
> there's no NaN nonsense to worry about), but maybe I'm missing something
> obvious.
>
> I've reproduced this on PostgreSQL 15.7 and 17beta3.
>
> Thanks for any insights y'all can provide!
>
> --
> James Brown
>

--00000000000050a7bd0620e79b78
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: base64
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=
--00000000000050a7bd0620e79b78--