postgresql 13 對於重複值的 index 優化

os: centos 7.4.1708
db: postgresql 13beta1
postgresql 12.3

版本

# cat /etc/centos-release
CentOS Linux release 7.4.1708 (Core) 
# 
# su - postgres
Last login: Fri Jun 19 10:46:55 CST 2020 on pts/0
$
$
$ psql -c "select version();"
                                                  version                                                   
------------------------------------------------------------------------------------------------------------
 PostgreSQL 13beta1 on x86_64-pc-linux-gnu, compiled by gcc (GCC) 4.8.5 20150623 (Red Hat 4.8.5-39), 64-bit
(1 row)

# cat /etc/centos-release
CentOS Linux release 7.4.1708 (Core) 
# 
# su - postgres
Last login: Wed Jan 15 18:34:12 CST 2020 on pts/0
$
$
$ psql -c "select version();"
                                                 version                                                 
---------------------------------------------------------------------------------------------------------
 PostgreSQL 12.3 on x86_64-pc-linux-gnu, compiled by gcc (GCC) 4.8.5 20150623 (Red Hat 4.8.5-39), 64-bit
(1 row)

創建索引

# su - postgres
$ psql

postgres=# create table tmp_t0(
id   int8,
name varchar(100),
memo varchar(100)
);

postgres=# insert into tmp_t0 
select '1',
       md5('1'::varchar),
       md5(id::varchar)	   
  from generate_series(1,1000000) as id;

postgres=# create index idx_tmp_t0_x1 on tmp_t0(name);
create index idx_tmp_t0_x2 on tmp_t0(memo);

postgres=# select * from tmp_t0 limit 10;
 id |               name               |               memo               
----+----------------------------------+----------------------------------
  1 | c4ca4238a0b923820dcc509a6f75849b | c4ca4238a0b923820dcc509a6f75849b
  1 | c4ca4238a0b923820dcc509a6f75849b | c81e728d9d4c2f636f067f89cc14862c
  1 | c4ca4238a0b923820dcc509a6f75849b | eccbc87e4b5ce2fe28308fd9f2a7baf3
  1 | c4ca4238a0b923820dcc509a6f75849b | a87ff679a2f3e71d9181a67b7542122c
  1 | c4ca4238a0b923820dcc509a6f75849b | e4da3b7fbbce2345d7772b0674a318d5
  1 | c4ca4238a0b923820dcc509a6f75849b | 1679091c5a880faf6fb5e6087eb1b2dc
  1 | c4ca4238a0b923820dcc509a6f75849b | 8f14e45fceea167a5a36dedd4bea2543
  1 | c4ca4238a0b923820dcc509a6f75849b | c9f0f895fb98ab9159f51fd0297e236d
  1 | c4ca4238a0b923820dcc509a6f75849b | 45c48cce2e2d7fbdea1afc51c7c6ad26
  1 | c4ca4238a0b923820dcc509a6f75849b | d3d9446802a44259755d38e6d163e820
(10 rows)

idx_tmp_t0_x1 爲值重複率很高的索引
idx_tmp_t0_x2 爲值重複率很低的索引

pg12 的兩個索引大小

postgres=# select pg_size_pretty(pg_relation_size('public.idx_tmp_t0_x1')),pg_size_pretty(pg_relation_size('public.idx_tmp_t0_x2'));
 pg_size_pretty | pg_size_pretty 
----------------+----------------
 56 MB          | 56 MB
(1 row)

pg13 的兩個索引大小

postgres=# select pg_size_pretty(pg_relation_size('public.idx_tmp_t0_x1')),pg_size_pretty(pg_relation_size('public.idx_tmp_t0_x2'));
 pg_size_pretty | pg_size_pretty 
----------------+----------------
 7152 kB        | 56 MB
(1 row)

pg12 idx_tmp_t0_x1的大小爲 56 MB
pg13 idx_tmp_t0_x1的大小爲 7152 KB

可以看到 pg13 針對有很多重複值的表列創建索引，大小的優化效果還是很明顯。

參考: