pg_hint_plan — tweak execution plans in SQL comments

pg_hint_plan makes it possible to tweak execution plans using "hints" in SQL comments, as of /*+ SeqScan(a) */.

ProtonBase uses a cost-based optimizer, which utilizes data statistics, not static rules. The planner (optimizer) estimates costs of each possible execution plans for a SQL statement then the execution plan with the lowest cost is executed. The planner does its best to select the best execution plan, but is not always perfect, since it doesn't take into account some of the data properties or correlations between columns.

Basic Usage

pg_hint_plan reads hinting phrases in a comment of special form given a SQL statement. A hint can be specified by prefixing it with the sequence "/\*+" and ending it with "\*/". Hint phrases consist of hint names and parameters enclosed by parentheses and delimited by whitespaces. Hint phrases can use newlines for readability.

In the example below, a hash join is selected as the join method while doing a sequential scan on pgbench_accounts:

=# /*+
     HashJoin(a b)
     SeqScan(a)
    */
   EXPLAIN SELECT *
     FROM pgbench_branches b
     JOIN pgbench_accounts a ON b.bid = a.bid
     ORDER BY a.aid;
                                        QUERY PLAN
---------------------------------------------------------------------------------------
    Sort  (cost=31465.84..31715.84 rows=100000 width=197)
    Sort Key: a.aid
    ->  <b>Hash Join</b>  (cost=1.02..4016.02 rows=100000 width=197)
            Hash Cond: (a.bid = b.bid)
            ->  <b>Seq Scan on pgbench_accounts a</b>  (cost=0.00..2640.00 rows=100000 width=97)
            ->  Hash  (cost=1.01..1.01 rows=1 width=100)
                ->  Seq Scan on pgbench_branches b  (cost=0.00..1.01 rows=1 width=100)
(7 rows)

The hint table

Hints can be specified in a comment, still this can be inconvenient in the case where queries cannot be edited. In the case, hints can be placed in a special table named "hint_plan.hints". The table consists of the following columns:

The following example shows how to operate with the hint table.

=# EXPLAIN (VERBOSE, COSTS false) SELECT * FROM t1 WHERE t1.id = 1;
               QUERY PLAN
----------------------------------------
 Seq Scan on public.t1
   Output: id, id2
   Filter: (t1.id = 1)
 Query Identifier: -7164653396197960701
(4 rows)
=# INSERT INTO hint_plan.hints(query_id, application_name, hints)
     VALUES (-7164653396197960701, '', 'SeqScan(t1)');
INSERT 0 1
=# UPDATE hint_plan.hints
     SET hints = 'IndexScan(t1)'
     WHERE id = 1;
UPDATE 1
=# DELETE FROM hint_plan.hints WHERE id = 1;
DELETE 1

The hint table is owned by the extension owner and has the same default privileges as of the time of its creation, during CREATE EXTENSION. Hints in the hint table are prioritized over hints in comments.

The query ID can be retrieved with pg_stat_statements or with EXPLAIN (VERBOSE).

Types of hints

Hinting phrases are classified in multiple types based on what kind of object and how they can affect the planner. See Hint list for more details.

Hints for Scan methods

Scan method hints enforce specific scanning methods on the target table. pg_hint_plan recognizes the target table by alias names if any. These are for example SeqScan or IndexScan.

Scan hints work on ordinary tables, inheritance tables, UNLOGGED tables, temporary tables and system catalogs. External (foreign) tables, table functions, VALUES clause, CTEs, views and subqueries are not affected.

=# /*+
     SeqScan(t1)
     IndexScan(t2 t2_pkey)
    */
   SELECT * FROM table1 t1 JOIN table table2 t2 ON (t1.key = t2.key);

Hints for Join methods

Join method hints enforce the join methods of the joins involving the specified tables.

This can affect joins only on ordinary tables. Inheritance tables, UNLOGGED tables, temporary tables, external (foreign) tables, system catalogs, table functions, VALUES command results and CTEs are allowed to be in the parameter list. Joins on views and subqueries are not affected.

Hints for Joining order

This hint, named "Leading", enforces the order of join on two or more tables. There are two methods of enforcing it. The first method enforces a specific order of joining but does not restrict the direction at each join level. The second method enforces the join direction additionally. See hint list for more details. For example:

=# /*+
     NestLoop(t1 t2)
     MergeJoin(t1 t2 t3)
     Leading(t1 t2 t3)
    */
   SELECT * FROM table1 t1
     JOIN table table2 t2 ON (t1.key = t2.key)
     JOIN table table3 t3 ON (t2.key = t3.key);

Hints for Row number corrections

This hint, named "Rows", changes the row number estimation of joins that comes from restrictions in the planner. For example:

=# /*+ Rows(a b #10) */ SELECT... ; Sets rows of join result to 10
=# /*+ Rows(a b +10) */ SELECT... ; Increments row number by 10
=# /*+ Rows(a b -10) */ SELECT... ; Subtracts 10 from the row number.
=# /*+ Rows(a b *10) */ SELECT... ; Makes the number 10 times larger.

Hints for parallel plans

This hint, named Parallel, enforces parallel execution configuration on scans. The third parameter specifies the strength of the enforcement. soft means that pg_hint_plan only changes max_parallel_worker_per_gather and leaves all the others to the planner to set. hard changes other planner parameters so as to forcibly apply the update. This can affect ordinary tables, inheritance parents, unlogged tables and system catalogs. External tables, table functions, VALUES clauses, CTEs, views and subqueries are not affected. Internal tables of a view can be specified by its real name or its alias as the target object. The following example shows that the query is enforced differently on each table:

=# EXPLAIN /*+ Parallel(c1 3 hard) Parallel(c2 5 hard) */
   SELECT c2.a FROM c1 JOIN c2 ON (c1.a = c2.a);
                                  QUERY PLAN
-------------------------------------------------------------------------------
 Hash Join  (cost=2.86..11406.38 rows=101 width=4)
   Hash Cond: (c1.a = c2.a)
   ->  Gather  (cost=0.00..7652.13 rows=1000101 width=4)
         Workers Planned: 3
         ->  Parallel Seq Scan on c1  (cost=0.00..7652.13 rows=322613 width=4)
   ->  Hash  (cost=1.59..1.59 rows=101 width=4)
         ->  Gather  (cost=0.00..1.59 rows=101 width=4)
               Workers Planned: 5
               ->  Parallel Seq Scan on c2  (cost=0.00..1.59 rows=59 width=4)
 
=# EXPLAIN /*+ Parallel(tl 5 hard) */ SELECT sum(a) FROM tl;
                                    QUERY PLAN
-----------------------------------------------------------------------------------
 Finalize Aggregate  (cost=693.02..693.03 rows=1 width=8)
   ->  Gather  (cost=693.00..693.01 rows=5 width=8)
         Workers Planned: 5
         ->  Partial Aggregate  (cost=693.00..693.01 rows=1 width=8)
               ->  Parallel Seq Scan on tl  (cost=0.00..643.00 rows=20000 width=4)

GUC parameters set during planning

Set hints change GUC parameters just while planning. GUC parameter shown in Query Planning can have the expected effects on planning unless an other hint conflicts with the planner method configuration parameters. When multiple hints change the same GUC, the last hint takes effect. GUC parameters for pg_hint_plan are also settable by this hint but it may not work as expected.

=# /*+ Set(random_page_cost 2.0) */
   SELECT * FROM table1 t1 WHERE key = 'value';
...

GUC parameters for pg_hint_plan

The following GUC parameters affect the behavior of pg_hint_plan:

Details in hinting

Syntax and placement

pg_hint_plan reads hints from only the first block comment and stops parsing from any characters except alphabetical characters, digits, spaces, underscores, commas and parentheses. In the following example, HashJoin(a b) and SeqScan(a) are parsed as hints, but IndexScan(a) and MergeJoin(a b) are not:

=# /*+
     HashJoin(a b)
     SeqScan(a)
    */
   /*+ IndexScan(a) */
   EXPLAIN SELECT /*+ MergeJoin(a b) */ *
     FROM pgbench_branches b
     JOIN pgbench_accounts a ON b.bid = a.bid
     ORDER BY a.aid;
                                      QUERY PLAN
---------------------------------------------------------------------------------------
 Sort  (cost=31465.84..31715.84 rows=100000 width=197)
   Sort Key: a.aid
   ->  Hash Join  (cost=1.02..4016.02 rows=100000 width=197)
         Hash Cond: (a.bid = b.bid)
         ->  Seq Scan on pgbench_accounts a  (cost=0.00..2640.00 rows=100000 width=97)
         ->  Hash  (cost=1.01..1.01 rows=1 width=100)
               ->  Seq Scan on pgbench_branches b  (cost=0.00..1.01 rows=1 width=100)
(7 rows)

Using with PL/pgSQL

pg_hint_plan works for queries in PL/pgSQL scripts with some restrictions.

• Hints affect only on the following kind of queries:
  • Queries that return one row (SELECT, INSERT, UPDATE and DELETE)
  • Queries that return multiple rows (RETURN QUERY)
  • Dynamic SQL statements (EXECUTE)
  • Cursor open (OPEN)
  • Loop over result of a query (FOR)
• A hint comment has to be placed after the first word in a query as preceding comments are not sent as a part of this query.

=# CREATE FUNCTION hints_func(integer) RETURNS integer AS $$
   DECLARE
     id  integer;
     cnt integer;
   BEGIN
     SELECT /*+ NoIndexScan(a) */ aid
       INTO id FROM pgbench_accounts a WHERE aid = $1;
     SELECT /*+ SeqScan(a) */ count(*)
       INTO cnt FROM pgbench_accounts a;
     RETURN id + cnt;
   END;
   $$ LANGUAGE plpgsql;

Upper and lower case handling in object names

Unlike the way ProtonBase handles object names, pg_hint_plan compares bare object names in hints against the database internal object names in a case-sensitive manner. Therefore, an object name TBL in a hint matches only "TBL" in the database and does not match any unquoted names like TBL, tbl or Tbl.

Escaping special characters in object names

The objects defined in a hint's parameter can use double quotes if they include parentheses, double quotes and white spaces. The escaping rules are the same as ProtonBase.

Distinction between multiple occurences of a table

pg_hint_plan identifies the target object by using aliases if any. This behavior is useful to point to a specific occurrence among multiple occurrences of one table.

=# /*+ HashJoin(t1 t1) */
   EXPLAIN SELECT * FROM s1.t1
     JOIN public.t1 ON (s1.t1.id=public.t1.id);
INFO:  hint syntax error at or near "HashJoin(t1 t1)"
DETAIL:  Relation name "t1" is ambiguous.
...
=# /*+ HashJoin(pt st) */
   EXPLAIN SELECT * FROM s1.t1 st
     JOIN public.t1 pt ON (st.id=pt.id);
                             QUERY PLAN
---------------------------------------------------------------------
 Hash Join  (cost=64.00..1112.00 rows=28800 width=8)
   Hash Cond: (st.id = pt.id)
   ->  Seq Scan on t1 st  (cost=0.00..34.00 rows=2400 width=4)
   ->  Hash  (cost=34.00..34.00 rows=2400 width=4)
         ->  Seq Scan on t1 pt  (cost=0.00..34.00 rows=2400 width=4)

Underlying tables of views or rules

Hints are not applicable on views, but they can affect the queries within the view if the object names match the names in the expanded query on the view. Assigning aliases to the tables in a view enables them to be manipulated from outside the view.

=# CREATE VIEW v1 AS SELECT * FROM t2;
=# EXPLAIN /*+ HashJoin(t1 v1) */
          SELECT * FROM t1 JOIN v1 ON (c1.a = v1.a);
                            QUERY PLAN
------------------------------------------------------------------
 Hash Join  (cost=3.27..18181.67 rows=101 width=8)
   Hash Cond: (t1.a = t2.a)
   ->  Seq Scan on t1  (cost=0.00..14427.01 rows=1000101 width=4)
   ->  Hash  (cost=2.01..2.01 rows=101 width=4)
         ->  Seq Scan on t2  (cost=0.00..2.01 rows=101 width=4)

Inheritance

Hints can only point to the parent of an inheritance tree and the hints affect all the tables in an inheritance tree. Hints pointing directly to inherited children have no effect.

Hints in multistatements

One multistatement can have exactly one hint comment and the hint affects all of the individual statements in the multistatement.

VALUES expressions

VALUES expressions in FROM clause are named as VALUES internally these can be hinted if it is the only VALUES of a query. Two or more VALUES expressions in a query cannot be distinguished by looking at an EXPLAIN result, resulting in ambiguous results:

=# /*+ MergeJoin(VALUES_1 VALUES) */
   EXPLAIN SELECT * FROM (VALUES (1, 1), (2, 2)) v (a, b)
     JOIN (VALUES (1, 5), (2, 8), (3, 4)) w (a, c) ON v.a = w.a;
INFO:  pg_hint_plan: hint syntax error at or near "MergeJoin(VALUES_1 VALUES) "
DETAIL:  Relation name "VALUES" is ambiguous.
                               QUERY PLAN
-------------------------------------------------------------------------
 Hash Join  (cost=0.05..0.12 rows=2 width=16)
   Hash Cond: ("VALUES_1".column1 = "VALUES".column1)
   ->  Values Scan on "VALUES_1"  (cost=0.00..0.04 rows=3 width=8)
   ->  Hash  (cost=0.03..0.03 rows=2 width=8)
         ->  Values Scan on "VALUES"  (cost=0.00..0.03 rows=2 width=8)

Subqueries

Subqueries context can be occasionally hinted using the name ANY_subquery:

IN (SELECT ... {LIMIT | OFFSET ...} ...)
= ANY (SELECT ... {LIMIT | OFFSET ...} ...)
= SOME (SELECT ... {LIMIT | OFFSET ...} ...)

For these syntaxes, the planner internally assigns the name to the subquery when planning joins on tables including it, so join hints are applicable on such joins using the implicit name. For example:

=# /*+HashJoin(a1 ANY_subquery)*/
   EXPLAIN SELECT *
     FROM pgbench_accounts a1
   WHERE aid IN (SELECT bid FROM pgbench_accounts a2 LIMIT 10);
                                         QUERY PLAN
 
---------------------------------------------------------------------------------------------
 Hash Semi Join  (cost=0.49..2903.00 rows=1 width=97)
   Hash Cond: (a1.aid = a2.bid)
   ->  Seq Scan on pgbench_accounts a1  (cost=0.00..2640.00 rows=100000 width=97)
   ->  Hash  (cost=0.36..0.36 rows=10 width=4)
         ->  Limit  (cost=0.00..0.26 rows=10 width=4)
               ->  Seq Scan on pgbench_accounts a2  (cost=0.00..2640.00 rows=100000 width=4)

Using IndexOnlyScan hint

Index scan may be unexpectedly performed on another index when the index specified in IndexOnlyScan hint cannot perform an index only scan.

About NoIndexScan

A NoIndexScan hint implies NoIndexOnlyScan.

Parallel hints and UNION

A UNION can run in parallel only when all underlying subqueries are parallel-safe. Hence, enforcing parallel on any of the subqueries will let a parallel-executable UNION run in parallel. Meanwhile, a parallel hint with zero workers prevents a scan from being executed in parallel.

Setting pg_hint_plan parameters by Set hints

pg_hint_plan parameters influence their own behavior so some parameters will not work as one could expect:

• Hints to change enable_hint, enable_hint_table are ignored even though they are reported as "used hints" in debug logs.
• Setting debug_print and message_level in the middle of query processing.

Functional limitations

Influence of planner GUC parameters

The planner does not try to consider joining order for FROM clause entries more than from_collapse_limit. pg_hint_plan cannot affect the joining order in this case.

Hints trying to enforce non-executable plans

Planner chooses any executable plans when the enforced plan cannot be executed:

• FULL OUTER JOIN to use nested loop.
• Use of indexes that do not have columns used in quals.
• TID scans for queries without ctid conditions.

Queries in ECPG

ECPG removes comments in queries written as embedded SQLs so hints cannot be passed to it. The only exception EXECUTE, that passes the query string to the server as-is. The hint table can be used in the case.

Query Identifiers

When compute_query_id is enabled, ProtonBase generates a query ID, ignoring comments. Hence, queries with different hints, still written the same way, may compute the same query ID.

Hint list

The available hints are listed below.