openGauss每日一练第19天|学习openGauss收集统计信息、打印执行计划、垃圾收集和checkpoint

学习目标

学习openGauss收集统计信息、打印执行计划、垃圾收集和checkpoint

课程学习

连接数据库

#第一次进入等待15秒
#数据库启动中...
su - omm
gsql -r

1.准备数据

Create schema tpcds;
CREATE TABLE tpcds.customer_address
(
ca_address_sk integer NOT NULL ,
ca_address_id character(16),
ca_street_number character(10) ,
ca_street_name character varying(60) ,
ca_street_type character(15) ,
ca_suite_number character(10) ,
ca_city character varying(60) ,
ca_county character varying(30) ,
ca_state character(2) ,
ca_zip character(10) ,
ca_country character varying(20) ,
ca_gmt_offset numeric(5,2) ,
ca_location_type character(20)
);
insert into tpcds.customer_address values
(1, 'AAAAAAAABAAAAAAA', '18', 'Jackson', 'Parkway', 'Suite 280', 'Fairfield', 'Maricopa County', 'AZ', '86192' ,'United States', -7.00, 'condo'),
(2, 'AAAAAAAACAAAAAAA', '362', 'Washington 6th', 'RD', 'Suite 80', 'Fairview', 'Taos County', 'NM', '85709', 'United States', -7.00, 'condo'),
(3, 'AAAAAAAADAAAAAAA', '585', 'Dogwood Washington', 'Circle', 'Suite Q', 'Pleasant Valley', 'York County', 'PA', '12477', 'United States', -5.00, 'single family');

–使用序列的generate_series(1,N)函数对表插入数据

insert into tpcds.customer_address values(generate_series(10, 10000));

2.收集统计信息

–查看系统表中表的统计信息

select relname, relpages, reltuples from pg_class where relname = 'customer_address';

—使用ANALYZE VERBOSE语句更新统计信息，并输出表的相关信息

analyze VERBOSE tpcds.customer_address;

–查看系统表中表的统计信息

select relname, relpages, reltuples from pg_class where relname = 'customer_address';

3.打印执行计划

–使用默认的打印格式

SET explain_perf_mode=normal;

–显示表简单查询的执行计划

EXPLAIN SELECT * FROM tpcds.customer_address;

–以JSON格式输出的执行计划（explain_perf_mode为normal时）

EXPLAIN(FORMAT JSON) SELECT * FROM tpcds.customer_address;

–禁止开销估计的执行计划

EXPLAIN(COSTS FALSE)SELECT * FROM tpcds.customer_address;

–带有聚集函数查询的执行计划

EXPLAIN SELECT SUM(ca_address_sk) FROM tpcds.customer_address WHERE ca_address_sk<100;

–有索引条件的执行计划

create index customer_address_idx on tpcds.customer_address(ca_address_sk);
EXPLAIN SELECT * FROM tpcds.customer_address WHERE ca_address_sk<100;

4.垃圾收集

–VACUUM回收表或B-Tree索引中已经删除的行所占据的存储空间

update tpcds.customer_address set ca_address_sk = ca_address_sk + 1 where ca_address_sk <100;
VACUUM (VERBOSE, ANALYZE) tpcds.customer_address;

5.事务日志检查点

–检查点（CHECKPOINT）是一个事务日志中的点，所有数据文件都在该点被更新以反映日志中的信息，所有数据文件都将被刷新到磁盘CHECKPOINT;

6.清理数据

drop schema tpcds cascade;

课后作业

1.创建分区表，并用generate_series(1,N)函数对表插入数据

omm=# Create schema tpcds;
CREATE SCHEMA
omm=# CREATE TABLE tpcds.customer_address
omm-# (
omm(#     ca_address_sk       integer                  NOT NULL   ,
omm(#     ca_address_id       character(16)                       ,
omm(#     ca_street_number    character(10)                       ,
omm(#     ca_street_name      character varying(60)               ,
omm(#     ca_street_type      character(15)                       ,
omm(#     ca_suite_number     character(10)                       ,
omm(# omm(#     ca_city             character varying(60)               ,
    ca_county           character varying(30)               ,
omm(#     ca_state            character(2)                        ,
omm(#     ca_zip              character(10)                       ,
omm(#     ca_country           character varying(20)               ,
omm(#     ca_gmt_offset       numeric(5,2)                        ,
omm(#     ca_location_type    character(20)
omm(# )    
omm-# PARTITION BY RANGE (ca_address_sk)
omm-# (
omm(#         PARTITION P1 VALUES LESS THAN(2000),
omm(#         PARTITION P2 VALUES LESS THAN(6000),
omm(#         PARTITION P3 VALUES LESS THAN(10000)
omm(# )
omm-# ENABLE ROW MOVEMENT;
CREATE TABLE

omm=# insert into tpcds.customer_address values(generate_series(10, 9999));
INSERT 0 9990

2.收集表统计信息

omm=# select relname, relpages, reltuples from pg_class where relname = 'customer_address';
     relname      | relpages | reltuples 
------------------+----------+-----------
 customer_address |        0 |         0
(1 row)

omm=# analyze VERBOSE tpcds.customer_address;
INFO:  analyzing "tpcds.customer_address"(gaussdb pid=1)
INFO:  ANALYZE INFO : "customer_address": scanned 22 of 22 pages, containing 1990 live rows and 1990 dead rows; 1990 rows in sample, 1990 estimated total rows(gaussdb pid=1)
INFO:  ANALYZE INFO : "customer_address": scanned 44 of 44 pages, containing 4000 live rows and 4000 dead rows; 4000 rows in sample, 4000 estimated total rows(gaussdb pid=1)
INFO:  ANALYZE INFO : "customer_address": scanned 44 of 44 pages, containing 4000 live rows and 4000 dead rows; 4000 rows in sample, 4000 estimated total rows(gaussdb pid=1)
ANALYZE
omm=# select relname, relpages, reltuples from pg_class where relname = 'customer_address';
     relname      | relpages | reltuples 
------------------+----------+-----------
 customer_address |      110 |      9990
(1 row)

3.显示简单查询的执行计划；建立索引并显示有索引条件的执行计划

omm=# SET explain_perf_mode=normal;
SET
omm=# EXPLAIN SELECT * FROM tpcds.customer_address;
                                       QUERY PLAN                                        
-----------------------------------------------------------------------------------------
 Partition Iterator  (cost=0.00..209.90 rows=9990 width=788)
   Iterations: 3
   ->  Partitioned Seq Scan on customer_address  (cost=0.00..209.90 rows=9990 width=788)
         Selected Partitions:  1..3
(4 rows)

omm=# create index customer_address_idx on tpcds.customer_address(ca_address_sk);
CREATE INDEX
omm=# EXPLAIN SELECT * FROM tpcds.customer_address WHERE ca_address_sk<100;
                                           QUERY PLAN                                           
------------------------------------------------------------------------------------------------
 Index Scan using customer_address_idx on customer_address  (cost=0.00..9.84 rows=91 width=788)
   Index Cond: (ca_address_sk < 100)
(2 rows)

4.更新表数据，并做垃圾收集

omm=# update tpcds.customer_address set ca_address_sk = ca_address_sk + 1 where ca_address_sk <100;
UPDATE 90
omm=# VACUUM (VERBOSE, ANALYZE) tpcds.customer_address;
INFO:  vacuuming "tpcds.customer_address"(gaussdb pid=1)
INFO:  scanned index "customer_address_idx" to remove 1990 row versions(gaussdb pid=1)
DETAIL:  CPU 0.00s/0.00u sec elapsed 0.00 sec.
DETAIL:  90 dead row versions cannot be removed yet.
There were 0 unused item pointers.
0 pages are entirely empty.
CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO:  index "customer_address_idx" now contains 2080 row versions in 31 pages(gaussdb pid=1)
DETAIL:  0 index row versions were removed.
0 index pages have been deleted, 0 are currently reusable.
CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO:  "customer_address": found 1990 removable, 2080 nonremovable row versions in 22 out of 22 pages(gaussdb pid=1)
INFO:  vacuuming "tpcds.customer_address"(gaussdb pid=1)
INFO:  scanned index "customer_address_idx" to remove 4000 row versions(gaussdb pid=1)
DETAIL:  CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO:  index "customer_address_idx" now contains 4000 row versions in 31 pages(gaussdb pid=1)
DETAIL:  0 index row versions were removed.
0 index pages have been deleted, 0 are currently reusable.
CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO:  "customer_address": found 4000 removable, 4000 nonremovable row versions in 44 out of 44 pages(gaussdb pid=1)
DETAIL:  0 dead row versions cannot be removed yet.
There were 0 unused item pointers.
0 pages are entirely empty.
CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO:  vacuuming "tpcds.customer_address"(gaussdb pid=1)
INFO:  scanned index "customer_address_idx" to remove 4000 row versions(gaussdb pid=1)
DETAIL:  CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO:  index "customer_address_idx" now contains 4000 row versions in 31 pages(gaussdb pid=1)
DETAIL:  0 index row versions were removed.
0 index pages have been deleted, 0 are currently reusable.
CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO:  "customer_address": found 4000 removable, 4000 nonremovable row versions in 44 out of 44 pages(gaussdb pid=1)
DETAIL:  0 dead row versions cannot be removed yet.
There were 0 unused item pointers.
0 pages are entirely empty.
CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO:  scanned index "customer_address_idx" to remove 0.000000 invisible rows(gaussdb pid=1)
DETAIL:  CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO:  analyzing "tpcds.customer_address"(gaussdb pid=1)
INFO:  ANALYZE INFO : "customer_address": scanned 22 of 22 pages, containing 1990 live rows and 90 dead rows; 1990 rows in sample, 1990 estimated total rows(gaussdb pid=1)
INFO:  ANALYZE INFO : "customer_address": scanned 44 of 44 pages, containing 4000 live rows and 0 dead rows; 4000 rows in sample, 4000 estimated total rows(gaussdb pid=1)
INFO:  ANALYZE INFO : "customer_address": scanned 44 of 44 pages, containing 4000 live rows and 0 dead rows; 4000 rows in sample, 4000 estimated total rows(gaussdb pid=1)
VACUUM

5.清理数据

omm=# drop schema tpcds cascade;
NOTICE:  drop cascades to table tpcds.customer_address
DROP SCHEMA