Implement table partitioning.

Table partitioning is like table inheritance and reuses much of the
existing infrastructure, but there are some important differences.
The parent is called a partitioned table and is always empty; it may
not have indexes or non-inherited constraints, since those make no
sense for a relation with no data of its own.  The children are called
partitions and contain all of the actual data.  Each partition has an
implicit partitioning constraint.  Multiple inheritance is not
allowed, and partitioning and inheritance can't be mixed.  Partitions
can't have extra columns and may not allow nulls unless the parent
does.  Tuples inserted into the parent are automatically routed to the
correct partition, so tuple-routing ON INSERT triggers are not needed.
Tuple routing isn't yet supported for partitions which are foreign
tables, and it doesn't handle updates that cross partition boundaries.

Currently, tables can be range-partitioned or list-partitioned.  List
partitioning is limited to a single column, but range partitioning can
involve multiple columns.  A partitioning "column" can be an
expression.

Because table partitioning is less general than table inheritance, it
is hoped that it will be easier to reason about properties of
partitions, and therefore that this will serve as a better foundation
for a variety of possible optimizations, including query planner
optimizations.  The tuple routing based which this patch does based on
the implicit partitioning constraints is an example of this, but it
seems likely that many other useful optimizations are also possible.

Amit Langote, reviewed and tested by Robert Haas, Ashutosh Bapat,
Amit Kapila, Rajkumar Raghuwanshi, Corey Huinker, Jaime Casanova,
Rushabh Lathia, Erik Rijkers, among others.  Minor revisions by me.
This commit is contained in:
Robert Haas
2016-12-07 13:17:43 -05:00
parent b7e1ae2328
commit f0e44751d7
85 changed files with 8896 additions and 281 deletions

View File

@ -536,3 +536,55 @@ FROM generate_series(1, 3) g(i);
reset enable_seqscan;
reset enable_indexscan;
reset enable_bitmapscan;
--
-- Check that constraint exclusion works correctly with partitions using
-- implicit constraints generated from the partition bound information.
--
create table list_parted (
a varchar
) partition by list (a);
create table part_ab_cd partition of list_parted for values in ('ab', 'cd');
create table part_ef_gh partition of list_parted for values in ('ef', 'gh');
create table part_null_xy partition of list_parted for values in (null, 'xy');
explain (costs off) select * from list_parted;
explain (costs off) select * from list_parted where a is null;
explain (costs off) select * from list_parted where a is not null;
explain (costs off) select * from list_parted where a in ('ab', 'cd', 'ef');
explain (costs off) select * from list_parted where a = 'ab' or a in (null, 'cd');
explain (costs off) select * from list_parted where a = 'ab';
create table range_list_parted (
a int,
b char(2)
) partition by range (a);
create table part_1_10 partition of range_list_parted for values from (1) to (10) partition by list (b);
create table part_1_10_ab partition of part_1_10 for values in ('ab');
create table part_1_10_cd partition of part_1_10 for values in ('cd');
create table part_10_20 partition of range_list_parted for values from (10) to (20) partition by list (b);
create table part_10_20_ab partition of part_10_20 for values in ('ab');
create table part_10_20_cd partition of part_10_20 for values in ('cd');
create table part_21_30 partition of range_list_parted for values from (21) to (30) partition by list (b);
create table part_21_30_ab partition of part_21_30 for values in ('ab');
create table part_21_30_cd partition of part_21_30 for values in ('cd');
create table part_40_inf partition of range_list_parted for values from (40) to (unbounded) partition by list (b);
create table part_40_inf_ab partition of part_40_inf for values in ('ab');
create table part_40_inf_cd partition of part_40_inf for values in ('cd');
create table part_40_inf_null partition of part_40_inf for values in (null);
explain (costs off) select * from range_list_parted;
explain (costs off) select * from range_list_parted where a = 5;
explain (costs off) select * from range_list_parted where b = 'ab';
explain (costs off) select * from range_list_parted where a between 3 and 23 and b in ('ab');
/* Should select no rows because range partition key cannot be null */
explain (costs off) select * from range_list_parted where a is null;
/* Should only select rows from the null-accepting partition */
explain (costs off) select * from range_list_parted where b is null;
explain (costs off) select * from range_list_parted where a is not null and a < 67;
explain (costs off) select * from range_list_parted where a >= 30;
drop table list_parted cascade;
drop table range_list_parted cascade;