本节简单介绍了PostgreSQL数据库中统计信息相关的数据字典表及其结构，包括pg_class、pg_statistic和pg_statistic_ext。
一、pg_class
在pg_class数据字典表中,存储了Relation(包括Index)的两个统计信息:页面占用总数relpages和元组总数reltuples,在优化器执行物理优化时用于估算访问路径的启动成本和总成本.
测试脚本如下,在t_grxx表中插入10w数据,创建2个索引:
drop table if exists t_grxx;
create table t_grxx(dwbh varchar(10),grbh varchar(10),xm varchar(20),xb varchar(10),nl int);
insert into t_grxx(dwbh,grbh,xm,xb,nl)
select generate_series(1,100000)/10||'',generate_series(1,100000),'XM'||generate_series(1,100000),
(case when (floor(random()*2)=0) then '男' else '女' end),floor(random() * 100 + 1)::int;
create index idx_t_grxx_grbh on t_grxx(grbh);
create index idx_t_grxx_dwbh on t_grxx(dwbh);
基础关系t_grxx中的统计信息
testdb=# select relpages,reltuples from pg_class where relname = 't_grxx';
relpages | reltuples
----------+-----------
726 | 100000 --> 页面数726,元组数10,000
(1 row)
索引idx_t_grxx_grbh的统计信息
testdb=# select relpages,reltuples from pg_class where relname = 'idx_t_grxx_grbh';
relpages | reltuples
----------+-----------
276 | 100000 --> 页面数276,元组数10,000
(1 row)
二、pg_statistic
pg_statistic是PG中存储统计信息的主要数据字典表.通过命令ANALYZE生成统计数据,这些统计数据在计划阶段提供给优化器使用,是成本估算的基础.
pg_statistic的表结构如下:
testdb=# \d pg_statistic
Table "pg_catalog.pg_statistic"
Column | Type | Collation | Nullable | Default
-------------+----------+-----------+----------+---------
starelid | oid | | not null |
staattnum | smallint | | not null |
stainherit | boolean | | not null |
stanullfrac | real | | not null |
stawidth | integer | | not null |
stadistinct | real | | not null |
stakind1 | smallint | | not null |
stakind2 | smallint | | not null |
stakind3 | smallint | | not null |
stakind4 | smallint | | not null |
stakind5 | smallint | | not null |
staop1 | oid | | not null |
staop2 | oid | | not null |
staop3 | oid | | not null |
staop4 | oid | | not null |
staop5 | oid | | not null |
stanumbers1 | real[] | | |
stanumbers2 | real[] | | |
stanumbers3 | real[] | | |
stanumbers4 | real[] | | |
stanumbers5 | real[] | | |
stavalues1 | anyarray | | |
stavalues2 | anyarray | | |
stavalues3 | anyarray | | |
stavalues4 | anyarray | | |
stavalues5 | anyarray | | |
Indexes:
"pg_statistic_relid_att_inh_index" UNIQUE, btree (starelid, staattnum, stainherit)
其中:
starelid:数据列所属的关系或索引Oid
staattnum:数据列编号
stainherit:是否继承表(分区表)
stanullfrac:NULL值所占的比例
stawidth:平均行大小,以字节为单位
stadistinct:非NULL值的唯一值信息.>0,表示多少个唯一值;=0,表示未知;<0,根据重复值出现次数求得:stadistinct = (-1)x(1-stanullfrac)/times(重复次数)
stakindN:在N号slot槽中存储的是哪种类型的统计信息
staopN:在N号槽中所使用的操作符,如"=","<",">"等
stanumbersN:在N号槽中存储的数值类型统计信息
stavaluesN*:在N号槽中存储的数据值(使用anyarray类型)
统计信息类型,在src/include/catalog/pg_statistic.h文件中有相关解释,定义如下:
#define STATISTIC_KIND_MCV 1
#define STATISTIC_KIND_HISTOGRAM 2
#define STATISTIC_KIND_CORRELATION 3
#define STATISTIC_KIND_MCELEM 4
#define STATISTIC_KIND_DECHIST 5
#define STATISTIC_KIND_RANGE_LENGTH_HISTOGRAM 6
#define STATISTIC_KIND_BOUNDS_HISTOGRAM 7
MCV=most common values,最常见值,即高频值
HISTOGRAM=HISTOGRAM,数据分布直方图
CORRELATION=CORRELATION,相关系数,该列未排序的数据分布与排序后的数据分布的相关性
MCELEM=most common elements,与MCV类似,存储的是最常见非NULL值
DECHIST=distinct elements count histogram,描述数组类型列的每行中不同值的分布。
RANGE_LENGTH_HISTOGRAM=length histogram,以行为单位的距离类型列的长度分布。
BOUNDS_HISTOGRAM=与HISTOGRAM类似,面向range-type column
仍以t_grxx表为例说明该表的统计信息
表结构:
testdb=# \d t_grxx
Table "public.t_grxx"
Column | Type | Collation | Nullable | Default
--------+-----------------------+-----------+----------+---------
dwbh | character varying(10) | | |
grbh | character varying(10) | | |
xm | character varying(20) | | |
xb | character varying(10) | | |
nl | integer | | |
Indexes:
"idx_t_grxx_dwbh" btree (dwbh)
"idx_t_grxx_grbh" btree (grbh)
数据列信息:
testdb=# select attrelid,attname,attnum from pg_attribute where attrelid = 16742 order by attnum;
attrelid | attname | attnum
----------+----------+--------
16742 | tableoid | -7
16742 | cmax | -6
16742 | xmax | -5
16742 | cmin | -4
16742 | xmin | -3
16742 | ctid | -1
16742 | dwbh | 1
16742 | grbh | 2
16742 | xm | 3
16742 | xb | 4
16742 | nl | 5
(11 rows)
dwbh/grbh/xm/xb/nl这5列的编号分别是1/2/3/4/5.
查询常规的统计信息
testdb=# select starelid,staattnum,stainherit,stanullfrac,stawidth,stadistinct from pg_statistic where starelid = 16742;
starelid | staattnum | stainherit | stanullfrac | stawidth | stadistinct
----------+-----------+------------+-------------+----------+-------------
16742 | 1 | f | 0 | 4 | 9984
16742 | 2 | f | 0 | 5 | -1
16742 | 3 | f | 0 | 7 | -1
16742 | 4 | f | 0 | 4 | 2
16742 | 5 | f | 0 | 4 | 100
(5 rows)
stanullfrac为0表示没有NULL值
stawidth表示各列的平均行大小,比如dwbh平均行大小为4个字节
stadistinct表示唯一值信息,比如dwbh,10000行中有9984个唯一值,而xb只有2个,grbh和xm则不存在重复值.
查询其他统计信息
testdb=# select starelid,staattnum,stakind1,stakind2,stakind3,stakind4,stakind5 from pg_statistic where starelid = 16742 order by staattnum;
starelid | staattnum | stakind1 | stakind2 | stakind3 | stakind4 | stakind5
----------+-----------+----------+----------+----------+----------+----------
16742 | 1 | 1 | 2 | 3 | 0 | 0
16742 | 2 | 2 | 3 | 0 | 0 | 0
16742 | 3 | 2 | 3 | 0 | 0 | 0
16742 | 4 | 1 | 3 | 0 | 0 | 0
16742 | 5 | 1 | 3 | 0 | 0 | 0
(5 rows)
第1列(dwbh),在第1/2/3个槽中分别存储了类型为第1/2/3种类型的统计信息,即STATISTIC_KIND_MCV/STATISTIC_KIND_HISTOGRAM/STATISTIC_KIND_CORRELATION,其他槽没有内容
第2列(grbh),在第1/2个槽中分别存储了类型为第2/3种类型的统计信息,即STATISTIC_KIND_HISTOGRAM/STATISTIC_KIND_CORRELATION,其他槽没有内容
第5列(nl),在第1/2个槽中分别存储了类型为第1/3种类型的统计信息,即STATISTIC_KIND_MCV/STATISTIC_KIND_CORRELATION,其他槽没有内容
第一列统计信息
testdb=# \x
Expanded display is on.
testdb=# select staattnum,stakind1,staop1,stanumbers1,stavalues1,
testdb-# stakind2,staop2,stanumbers2,stavalues2,
testdb-# stakind3,staop3,stanumbers3,stavalues3
testdb-# from pg_statistic
testdb-# where starelid = 16742
testdb-# and staattnum = 1;
-[ RECORD 1 ]---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
staattnum | 1 -->dwbh
stakind1 | 1 -->STATISTIC_KIND_MCV,高频值统计信息
staop1 | 98 -->"=",查询pg_operator,select * from pg_operator where oid=98;
stanumbers1 | {0.0003} -->高频值比例,即3/10000
stavalues1 | {24} -->表示'24'这个值
stakind2 | 2 -->STATISTIC_KIND_HISTOGRAM,直方图
staop2 | 664 -->"<",左闭右开的区间
stanumbers2 | -->无
stavalues2 | {0,1084,1172,1262,1356,1441,1533,1622,1703,1790,1885,198,207,2165,2259,2351,2447,2530,262,2708,2799,2895,2985,3087,3172,3262,3359,3459,3541,3629,3716,3805,3900,3995,4078,417,4257,4345,4432,4515,461,47,479,489,4985,5069,5154,5244,533,542,5510,5596,5686,5786,587,5963,6053,6145,6232,6323,6412,650,6599,6686,6779,6868,6957,7038,7127,7218,7305,7403,7495,7588,7679,7767,7857,7942,8037,8125,8209,8307,8392,8481,8575,8664,8755,8844,8935,9022,9115,9202,9296,9378,9464,9561,965,9731,982,9906,9999}
-->剔除高频值后的数据分布,假定平均分布,0为MIN值,9999为MAX值
-->[0,1084),[1084,1172),...[9906,9999)
stakind3 | 3 -->STATISTIC_KIND_CORRELATION,相关系数
staop3 | 664 -->"<"
stanumbers3 | {0.817163}-->未排序和已排序的数据分布,有81.7%的相关性,值越高,顺序扫描结果越近似于排序
stavalues3 | -->无
第二列统计信息
testdb=# \x
Expanded display is on.
testdb=# select staattnum,stakind1,staop1,stanumbers1,stavalues1,
stakind2,staop2,stanumbers2,stavalues2,
stakind3,staop3,stanumbers3,stavalues3
from pg_statistic
where starelid = 16742
and staattnum = 2;
-[ RECORD 1 ]---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
staattnum | 2 -->grbh
stakind1 | 2 -->STATISTIC_KIND_HISTOGRAM,直方图
staop1 | 664 -->"<"
stanumbers1 | -->NULL
stavalues1 | {1,1088,11753,12659,1358,14448,15354,16253,17071,17939,18872,19821,20730,21676,22630,23540,24479,25302,26198,27069,27974,2894,29836,30851,31725,32619,33587,34584,35427,36280,37164,3806,39000,39942,40786,41695,42571,43452,44330,4516,46102,46997,47900,48899,49852,50697,5157,52452,533,54204,5512,55963,5686,57860,5871,59633,60544,61468,62331,63252,64134,65016,65993,66870,67798,68687,69568,70391,71285,72205,73066,74045,74962,75881,7679,77675,78573,79434,8038,81276,82125,83078,83939,84822,85764,86654,87557,88450,89352,90249,91171,92030,92979,93797,94665,95628,96510,97328,98212,99074,99998}
-->参照第一列
stakind2 | 3 -->STATISTIC_KIND_CORRELATION,相关系数
staop2 | 664 -->"<"
stanumbers2 | {0.816172}-->未排序和已排序的数据分布
stavalues2 | -->NULL
stakind3 | 0 -->第3个槽无统计信息
staop3 | 0
stanumbers3 |
stavalues3 |
第五列统计信息
testdb=# select staattnum,stakind1,staop1,stanumbers1,stavalues1,
testdb-# stakind2,staop2,stanumbers2,stavalues2,
testdb-# stakind3,staop3,stanumbers3,stavalues3
testdb-# from pg_statistic
testdb-# where starelid = 16742
testdb-# and staattnum = 5;
-[ RECORD 1 ]-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
staattnum | 5 -->nl
stakind1 | 1 -->STATISTIC_KIND_MCV,高频值统计信息
staop1 | 96 -->"="
stanumbers1 | {0.0117667,0.0117333,0.0113,0.0112667,0.0112333,0.0110333,0.0109333,0.0109333,0.0109333,0.0108333,0.0108,0.0108,0.0107667,0.0107667,0.0107333,0.0107333,0.0107333,0.0107,0.0106333,0.0106333,0.0106333,0.0105667,0.0105333,0.0105333,0.0105,0.0104667,0.0104667,0.0104333,0.0104333,0.0103667,0.0103,0.0102667,0.0102333,0.0102333,0.0102333,0.0102333,0.0102,0.0102,0.0102,0.0101667,0.0101667,0.0101333,0.0101,0.0101,0.0100667,0.0100333,0.0100333,0.01,0.00996667,0.00996667,0.00996667,0.00996667,0.00993333,0.00993333,0.00993333,0.00993333,0.0099,0.00986667,0.00983333,0.00983333,0.0098,0.0098,0.0098,0.00973333,0.0097,0.00966667,0.00966667,0.00963333,0.00963333,0.00963333,0.00963333,0.0096,0.0096,0.00956667,0.00953333,0.0095,0.0095,0.0095,0.0095,0.00943333,0.0094,0.00936667,0.00936667,0.00933333,0.00933333,0.0093,0.00916667,0.00916667,0.00916667,0.00913333,0.0091,0.0091,0.0091,0.00906667,0.009,0.00896667,0.00893333,0.00876667,0.00876667,0.0079}
-->每个高频值的出现频率
stavalues1 | {73,40,7,51,15,8,14,74,100,53,77,89,33,42,26,29,78,11,16,38,58,18,27,95,49,34,92,65,87,30,32,28,23,62,82,96,1,50,60,36,70,52,9,37,84,54,61,5,59,72,80,93,35,66,79,81,55,68,88,98,13,47,56,22,24,39,43,3,21,75,85,25,44,71,86,4,20,83,97,91,99,63,94,17,46,57,48,67,69,90,2,6,45,19,10,64,76,31,41,12}
-->高频值
stakind2 | 3 -->STATISTIC_KIND_CORRELATION,相关系数
staop2 | 97 -->"<",整型比较,664是字符型比较
stanumbers2 | {0.00562935} -->0.5%的相关性,相关性较低
stavalues2 | -->NULL
stakind3 | 0 -->无相关信息
staop3 | 0
stanumbers3 |
stavalues3 |
三、pg_statistic_ext
pg_statistic_ext数据字典表用于存储多列统计信息,需使用CREATE STATISTICS命令创建统计信息,在执行ANALYZE命令时统计.
仍以t_grxx为例,统计dwbh和grbh这两列的信息
testdb=# create statistics sta_t_grxx_dwbh_grbh on dwbh,grbh from t_grxx;
CREATE STATISTICS
testdb=# analyze t_grxx(dwbh,grbh);
ANALYZE
查询多列统计信息
testdb=# \x
Expanded display is on.
testdb=# select * from pg_statistic_ext;
-[ RECORD 1 ]---+-----------------------------------------
stxrelid | 16742 -->数据表Oid
stxname | sta_t_grxx_dwbh_grbh -->统计信息名称
stxnamespace | 2200 -->表空间
stxowner | 10 -->统计信息的Owner
stxkeys | 1 2 -->列编号,1 2表示dwbh grbh
stxkind | {d,f} -->STATS_EXT_NDISTINCT/STATS_EXT_DEPENDENCIES
stxndistinct | {"1, 2": 100000} -->STATS_EXT_NDISTINCT统计信息
stxdependencies | {"1 => 2": 0.040900, "2 => 1": 1.000000} -->STATS_EXT_DEPENDENCIES统计信息
四、参考资料
pg_statistic
pg_statistic.h
CREATE STATISTICS