本节介绍了PostgreSQL创建数据表时的参数fillfactor,该参数限制了插入数据时序预留多少空闲空间(比例),对于数据表来说默认值为100,索引是90,其解释如下: table >The fillfactor for a table is a percentage between 10 and 100. 100 (complete packing) is the default. When a smaller fillfactor is specified, INSERT operations pack table pages only to the indicated percentage; the remaining space on each page is reserved for updating rows on that page. This gives UPDATE a chance to place the updated copy of a row on the same page as the original, which is more efficient than placing it on a different page. For a table whose entries are never updated, complete packing is the best choice, but in heavily updated tables smaller fillfactors are appropriate. index >The fillfactor for an index is a percentage that determines how full the index method will try to pack index pages. For B-trees, leaf pages are filled to this percentage during initial index build, and also when extending the index at the right (largest key values). If pages subsequently become completely full, they will be split, leading to gradual degradation in the index's efficiency. B-trees use a default fillfactor of 90, but any value from 10 to 100 can be selected. If the table is static then fillfactor 100 is best to minimize the index's physical size, but for heavily updated tables a smaller fillfactor is better to minimize the need for page splits. The other index methods use fillfactor in different but roughly analogous ways; the default fillfactor varies between methods. 下面创建不同fillfactor的数据表,执行update操作 ``` [local]:5432 pg12@testdb=# create table t_fillfactor_100(id int,c1 varchar(30),c2 varchar(30),c3 varchar(30)) with (fillfactor=100); CREATE TABLE Time: 2.462 ms [local]:5432 pg12@testdb=# create table t_fillfactor_70(id int,c1 varchar(30),c2 varchar(30),c3 varchar(30)) with (fillfactor=70); CREATE TABLE Time: 3.437 ms [local]:5432 pg12@testdb=# create table t_fillfactor_50(id int,c1 varchar(30),c2 varchar(30),c3 varchar(30)) with (fillfactor=50); CREATE TABLE Time: 28.553 ms [local]:5432 pg12@testdb=# insert into t_fillfactor_100(id,c1,c2,c3) select x,'c1'||x,'c2'||x,'c3'||x from generate_series(1,1000000) as x; INSERT 0 1000000 Time: 3583.216 ms (00:03.583) [local]:5432 pg12@testdb=# insert into t_fillfactor_70(id,c1,c2,c3) select x,'c1'||x,'c2'||x,'c3'||x from generate_series(1,1000000) as x; INSERT 0 1000000 Time: 6506.113 ms (00:06.506) [local]:5432 pg12@testdb=# insert into t_fillfactor_50(id,c1,c2,c3) select x,'c1'||x,'c2'||x,'c3'||x from generate_series(1,1000000) as x; INSERT 0 1000000 Time: 3113.901 ms (00:03.114) [local]:5432 pg12@testdb=# update t_fillfactor_100 set c1=lpad('c1',30,'c1'),c2=lpad('c2',30,'c2'),c3=lpad('c3',30,'c3'); UPDATE 1000000 Time: 10641.794 ms (00:10.642) [local]:5432 pg12@testdb=# update t_fillfactor_70 set c1=lpad('c1',30,'c1'),c2=lpad('c2',30,'c2'),c3=lpad('c3',30,'c3'); UPDATE 1000000 Time: 8563.046 ms (00:08.563) [local]:5432 pg12@testdb=# update t_fillfactor_50 set c1=lpad('c1',30,'c1'),c2=lpad('c2',30,'c2'),c3=lpad('c3',30,'c3'); UPDATE 1000000 Time: 4036.735 ms (00:04.037) ``` 可以看到,在插入时,fillfactor较高的数据表耗时较短,而在update时(全量),fillfactor的则有较大的优势.但,经过多次update后,耗时并不明显,原因在于经过多次update,每个块的空闲空间跟fillfactor=100的设定已相差无几. ``` [local]:5432 pg12@testdb=# update t_fillfactor_100 set c1=lpad('c1',30,'c1'),c2=lpad('c2',30,'c2'),c3=lpad('c3',30,'c3'); UPDATE 1000000 Time: 4276.404 ms (00:04.276) [local]:5432 pg12@testdb=# update t_fillfactor_70 set c1=lpad('c1',30,'c1'),c2=lpad('c2',30,'c2'),c3=lpad('c3',30,'c3'); UPDATE 1000000 Time: 3856.575 ms (00:03.857) [local]:5432 pg12@testdb=# update t_fillfactor_50 set c1=lpad('c1',30,'c1'),c2=lpad('c2',30,'c2'),c3=lpad('c3',30,'c3'); UPDATE 1000000 Time: 4364.962 ms (00:04.365) [local]:5432 pg12@testdb=# ``` 重新创建表,使用pgbench进行测试 ``` [local]:5432 pg12@testdb=# drop table if exists t_fillfactor_100; t,c1 varchar(30),c2 varchar(30),c3 varchar(30)) with (fillfactor=70); create table t_fillfactor_50(id int,c1 varchar(30),c2 varchar(30),c3 varchar(30)) with (fillfactor=50); insert into t_fillfactor_100(id,c1,c2,c3) select x,'c1'||x,'c2'||x,'c3'||x from generate_series(1,1000000) as x; insert into t_fillfactor_70(id,c1,c2,c3) select x,'c1'||x,'c2'||x,'c3'||x from generate_series(1,1000000) as x; insert into t_fillfactor_50(id,c1,c2,c3) select x,'c1'||x,'c2'||x,'c3'||x from generate_series(1,1000000) as x; DROP TABLE Time: 191.706 ms [local]:5432 pg12@testdb=# drop table if exists t_fillfactor_70; DROP TABLE Time: 35.313 ms [local]:5432 pg12@testdb=# drop table if exists t_fillfactor_50; DROP TABLE Time: 30.078 ms [local]:5432 pg12@testdb=# [local]:5432 pg12@testdb=# create table t_fillfactor_100(id int,c1 varchar(30),c2 varchar(30),c3 varchar(30)) with (fillfactor=100); CREATE TABLE Time: 40.443 ms [local]:5432 pg12@testdb=# create table t_fillfactor_70(id int,c1 varchar(30),c2 varchar(30),c3 varchar(30)) with (fillfactor=70); CREATE TABLE Time: 1.334 ms [local]:5432 pg12@testdb=# create table t_fillfactor_50(id int,c1 varchar(30),c2 varchar(30),c3 varchar(30)) with (fillfactor=50); CREATE TABLE Time: 1.024 ms [local]:5432 pg12@testdb=# [local]:5432 pg12@testdb=# insert into t_fillfactor_100(id,c1,c2,c3) select x,'c1'||x,'c2'||x,'c3'||x from generate_series(1,1000000) as x; INSERT 0 1000000 Time: 2623.943 ms (00:02.624) [local]:5432 pg12@testdb=# insert into t_fillfactor_70(id,c1,c2,c3) select x,'c1'||x,'c2'||x,'c3'||x from generate_series(1,1000000) as x; INSERT 0 1000000 Time: 2543.045 ms (00:02.543) [local]:5432 pg12@testdb=# insert into t_fillfactor_50(id,c1,c2,c3) select x,'c1'||x,'c2'||x,'c3'||x from generate_series(1,1000000) as x; INSERT 0 1000000 Time: 2662.223 ms (00:02.662) [local]:5432 pg12@testdb=# ``` 使用pgbench进行测试 ``` [pg12@localhost script]$ cat update_100.sql \set id random(1,1000000) begin; update t_fillfactor_100 set c1=lpad('c1',30,'c1'),c2=lpad('c2',30,'c2'),c3=lpad('c3',30,'c3') where id= :id; end; [pg12@localhost script]$ cat update_70.sql \set id random(1,1000000) begin; update t_fillfactor_70 set c1=lpad('c1',30,'c1'),c2=lpad('c2',30,'c2'),c3=lpad('c3',30,'c3') where id= :id; end; [pg12@localhost script]$ cat update_50.sql \set id random(1,1000000) begin; update t_fillfactor_50 set c1=lpad('c1',30,'c1'),c2=lpad('c2',30,'c2'),c3=lpad('c3',30,'c3') where id= :id; end; [pg12@localhost script]$ pgbench -c 2 -C -f ~/script/update_100.sql -j 1 -n -T 60 -U pg12 testdb transaction type: /home/pg12/script/update_100.sql scaling factor: 1 query mode: simple number of clients: 2 number of threads: 1 duration: 60 s number of transactions actually processed: 691 latency average = 174.136 ms tps = 11.485277 (including connections establishing) tps = 11.625959 (excluding connections establishing) [pg12@localhost script]$ pgbench -c 2 -C -f ~/script/update_70.sql -j 1 -n -T 60 -U pg12 testdb transaction type: /home/pg12/script/update_70.sql scaling factor: 1 query mode: simple number of clients: 2 number of threads: 1 duration: 60 s number of transactions actually processed: 652 latency average = 184.293 ms tps = 10.852275 (including connections establishing) tps = 10.981136 (excluding connections establishing) [pg12@localhost script]$ pgbench -c 2 -C -f ~/script/update_50.sql -j 1 -n -T 60 -U pg12 testdb transaction type: /home/pg12/script/update_50.sql scaling factor: 1 query mode: simple number of clients: 2 number of threads: 1 duration: 60 s number of transactions actually processed: 627 latency average = 191.700 ms tps = 10.432967 (including connections establishing) tps = 10.551899 (excluding connections establishing) [pg12@localhost script]$ ``` 使用pgbench使用随机值进行测试,时长为60s,结果差别不大. 数据表大小的比较： ``` [local]:5432 pg12@testdb=# select pg_size_pretty(pg_relation_size('t_fillfactor_100')); pg_size_pretty ---------------- 58 MB (1 row) Time: 2.034 ms [local]:5432 pg12@testdb=# select pg_size_pretty(pg_relation_size('t_fillfactor_70')); pg_size_pretty ---------------- 82 MB (1 row) Time: 1.469 ms [local]:5432 pg12@testdb=# select pg_size_pretty(pg_relation_size('t_fillfactor_50')); pg_size_pretty ---------------- 117 MB (1 row) Time: 2.531 ms [local]:5432 pg12@testdb=# ``` 分别是58MB vs 82MB vs 117MB ≈ 100 vs 70 vs 50 **参考资料** [difference-between-table-fillfactor-and-index-fillfactor](https://dba.stackexchange.com/questions/15306/difference-between-table-fillfactor-and-index-fillfactor/15307 )