标签：sql调优集
************************************************************
1.新建调优集对象
************************************************************
---授权
grant ADMINISTER ANY SQL TUNING SET to scott;
---删除存在的STS
BEGIN
DBMS_SQLTUNE.DROP_SQLSET(
sqlset_name => 'OCPYANG_STS'
);
END;
/
---新建STS
BEGIN
DBMS_SQLTUNE.CREATE_SQLSET(
sqlset_name => 'OCPYANG_STS',
sqlset_owner => 'SCOTT',
description => 'ocpyangtest');
END;
/
---查看数据库已经创建的SQLSET
select owner, name, id, created, statement_count from dba_sqlset;
************************************************************
2.查看AWR资源密集型SQL语句
************************************************************
---2.1查看可用的快照范围
SELECT snap_id, instance_number, end_interval_time
FROM dba_hist_snapshot
ORDER BY snap_id;
---2.2 查看快照编号820-840之间磁盘使用率前10的sql
SELECT sql_id
,substr(sql_text,1,100)
,disk_reads, cpu_time, elapsed_time
FROM table(DBMS_SQLTUNE.SELECT_WORKLOAD_REPOSITORY(820,841,
null, null, 'disk_reads',null, null, null, 10))
ORDER BY disk_reads DESC;
---2.3查看没有被sys用户解析的sql
SELECT sql_id, substr(sql_text,1,100)
,disk_reads, cpu_time, elapsed_time, parsing_schema_name
FROM table(
DBMS_SQLTUNE.SELECT_WORKLOAD_REPOSITORY(820,841,
'parsing_schema_name <> ''SYS''',
NULL, NULL,NULL,NULL, 1, NULL, 'ALL'));
---2.4 查看快照编号820-840之间非sql用户排序的前10的sql
SELECT sql_id, substr(sql_text,1,100)
,disk_reads, cpu_time, elapsed_time, buffer_gets, parsing_schema_name
FROM table(
DBMS_SQLTUNE.SELECT_WORKLOAD_REPOSITORY(
begin_snap => 820
,end_snap => 841
,basic_filter => 'parsing_schema_name <> ''SYS'''
,ranking_measure1 => 'buffer_gets'
,result_limit => 10
));
COL bsnap NEW_VALUE begin_snap
COL esnap NEW_VALUE end_snap
--
SELECT MAX(snap_id) bsnap
FROM dba_hist_snapshot
WHERE begin_interval_time < sysdate-7;
--
SELECT MAX(snap_id) esnap
FROM dba_hist_snapshot;
--
COL sql_text FORMAT A40
COL sql_id FORMAT A15
COL parsing_schema_name FORMAT A15
COL cpu_seconds FORMAT 999,999,999,999,999
SET LONG 10000 LINES 132 PAGES 100 TRIMSPOOL ON
--
SELECT sql_id, sql_text
,disk_reads, cpu_time cpu_seconds, elapsed_time, buffer_gets, parsing_schema_name
FROM table(
DBMS_SQLTUNE.SELECT_WORKLOAD_REPOSITORY(
begin_snap => &begin_snap
,end_snap => &end_snap
,basic_filter => 'parsing_schema_name <> ''SYS'''
,ranking_measure1 => 'cpu_time'
,result_limit => 10
));
************************************************************
3.使用AWR中高资源消耗的SQL来填充优化集:
************************************************************
---3.1新建STS
BEGIN
DBMS_SQLTUNE.CREATE_SQLSET(
sqlset_name => 'OCPYANG_STS',
sqlset_owner => 'SCOTT',
description => 'ocpyangtest');
END;
/
---3.2查看AWR快照起止
select snap_id, begin_interval_time
from dba_hist_snapshot order by 1;
---3.3使用AWR中高资源的sql来填充sql优化集
DECLARE
test_cur dbms_sqltune.sqlset_cursor;
BEGIN
OPEN test_cur FOR
SELECT value(x)
FROM table(dbms_sqltune.select_workload_repository(
820,841, null, null,'disk_reads',
null, null, null, 15)) x;
--
dbms_sqltune.load_sqlset(
sqlset_owner =>'SCOTT',
sqlset_name => 'OCPYANG_STS',
populate_cursor => test_cur);
END;
/
*------------------常见错误
第 1 行出现错误:
ORA-13774: 权限不足, 无法从工作量资料档案库中选择数据 ORA-06512: 在
"SYS.DBMS_SQLTUNE", line 4715
ORA-06512: 在 line 10
使用sys账户即可，在DBMS_SQLTUNE.LOAD_SQLSET指定sqlset_owner
DBMS_SQLTUNE.LOAD_SQLSET (
sqlset_name IN VARCHAR2,
populate_cursor IN sqlset_cursor,
load_option IN VARCHAR2 := 'INSERT',
update_option IN VARCHAR2 := 'REPLACE',
update_condition IN VARCHAR2 := NULL,
update_attributes IN VARCHAR2 := NULL,
ignore_null IN BOOLEAN := TRUE,
commit_rows IN POSITIVE := NULL,
sqlset_owner IN VARCHAR2 := NULL);
-----------------------------*
---3.4 查看优化集相信信息
SELECT sqlset_name, elapsed_time
,cpu_time, buffer_gets, disk_reads, sql_text
FROM dba_sqlset_statements
WHERE sqlset_name = 'OCPYANG_STS';
************************************************************
4.查看内存中资源密集型的sql
************************************************************
---4.1 语法
DBMS_SQLTUNE.SELECT_CURSOR_CACHE (
basic_filter IN VARCHAR2 := NULL,
object_filter IN VARCHAR2 := NULL,
ranking_measure1 IN VARCHAR2 := NULL,
ranking_measure2 IN VARCHAR2 := NULL,
ranking_measure3 IN VARCHAR2 := NULL,
result_percentage IN NUMBER := 1,
result_limit IN NUMBER := NULL,
attribute_list IN VARCHAR2 := NULL)
RETURN sys.sqlset PIPELINED;
---4.2 从内存中选择读取磁盘超过1000000
SELECT sql_id, substr(sql_text,1,20), disk_reads
,cpu_time, elapsed_time
,buffer_gets, parsing_schema_name
FROM table(DBMS_SQLTUNE.SELECT_CURSOR_CACHE('disk_reads>1000000')
)
order by sql_id;
---4.3 查看内存中非sys账户用户CPU时间最长的10个查询
SELECT sql_id, substr(sql_text,1,120), disk_reads
,cpu_time, elapsed_time
,buffer_gets, parsing_schema_name
FROM table(
DBMS_SQLTUNE.SELECT_CURSOR_CACHE(
basic_filter => 'parsing_schema_name <> ''SYS'''
,ranking_measure1 => 'cpu_time'
,result_limit => 10
));
---4.4 查看内存中非sys账户运行返回时间超过1秒的
SELECT sql_id, substr(sql_text,1,120)
,disk_reads, cpu_time, elapsed_time
FROM table(DBMS_SQLTUNE.SELECT_CURSOR_CACHE('parsing_schema_name <> ''SYS''
AND elapsed_time > 1000000'))
ORDER BY sql_id;
---4.5 查看具体sql_id执行细节
SELECT *
FROM table(DBMS_SQLTUNE.SELECT_CURSOR_CACHE('sql_id = ''byzwu34haqkn4'''));
----4.6 各种案例
-- Select all statements in the cursor cache.
DECLARE
cur sys_refcursor;
BEGIN
OPEN cur FOR
SELECT value(P)
FROM table(DBMS_SQLTUNE.SELECT_CURSOR_CACHE) P;
-- Process each statement (or pass cursor to load_sqlset).
CLOSE cur;
END;/
-- Look for statements not parsed by SYS.
DECLARE
cur sys_refcursor;
BEGIN
OPEN cur for
SELECT VALUE(P)
FROM table(
DBMS_SQLTUNE.SELECT_CURSOR_CACHE('parsing_schema_name <> ''SYS''')) P;
-- Process each statement (or pass cursor to load_sqlset).
CLOSE cur;
end;/
-- All statements from a particular module/action.
DECLARE
cur sys_refcursor;
BEGIN
OPEN cur FOR
SELECT VALUE(P)
FROM table(
DBMS_SQLTUNE.SELECT_CURSOR_CACHE(
'module = ''MY_APPLICATION'' and action = ''MY_ACTION''')) P;
-- Process each statement (or pass cursor to load_sqlset)
CLOSE cur;
END;/
-- all statements that ran for at least five seconds
DECLARE
cur sys_refcursor;
BEGIN
OPEN cur FOR
SELECT VALUE(P)
FROM table(DBMS_SQLTUNE.SELECT_CURSOR_CACHE('elapsed_time > 5000000')) P;
-- Process each statement (or pass cursor to load_sqlset)
CLOSE cur;
end;/
-- select all statements that pass a simple buffer_gets threshold and
-- are coming from an APPS user
DECLARE
cur sys_refcursor;
BEGIN
OPEN cur FOR
SELECT VALUE(P)
FROM table(
DBMS_SQLTUNE.SELECT_CURSOR_CACHE(
'buffer_gets > 100 and parsing_schema_name = ''APPS'''))P;
-- Process each statement (or pass cursor to load_sqlset)
CLOSE cur;
end;/
-- select all statements exceeding 5 seconds in elapsed time, but also
-- select the plans (by default we only select execution stats and binds
-- for performance reasons - in this case the SQL_PLAN attribute of sqlset_row
-- is NULL)
DECLARE
cur sys_refcursor;
BEGIN
OPEN cur FOR
SELECT VALUE(P)
FROM table(dbms_sqltune.select_cursor_cache(
'elapsed_time > 5000000', NULL, NULL, NULL, NULL, 1, NULL,
'EXECUTION_STATISTICS, SQL_BINDS, SQL_PLAN')) P;
-- Process each statement (or pass cursor to load_sqlset)
CLOSE cur;
END;/
-- Select the top 100 statements in the cursor cache ordering by elapsed_time.
DECLARE
cur sys_refcursor;
BEGIN
OPEN cur FOR
SELECT VALUE(P)
FROM table(DBMS_SQLTUNE.SELECT_CURSOR_CACHE(NULL,
NULL,
'ELAPSED_TIME', NULL, NULL,
1,
100)) P;
-- Process each statement (or pass cursor to load_sqlset)
CLOSE cur;
end;/
-- Select the set of statements which cumulatively account for 90% of the
-- buffer gets in the cursor cache. This means that the buffer gets of all
-- of these statements added up is approximately 90% of the sum of all
-- statements currently in the cache.
DECLARE
cur sys_refcursor;
BEGIN
OPEN cur FOR
SELECT VALUE(P)
FROM table(DBMS_SQLTUNE.SELECT_CURSOR_CACHE(NULL,
NULL,
'BUFFER_GETS', NULL, NULL,
.9)) P;
-- Process each statement (or pass cursor to load_sqlset).
CLOSE cur;
END;
/
************************************************************
5.用内存中高资源消耗的sql填充调优集
************************************************************
---5.0 删除存在的STS
BEGIN
DBMS_SQLTUNE.DROP_SQLSET(
sqlset_name => 'OCPYANG_STS'
);
END;
/
--5.1新建调优集
BEGIN
DBMS_SQLTUNE.CREATE_SQLSET(
sqlset_name => 'OCPYANG_STS',
sqlset_owner => 'SCOTT',
description => 'ocpyangtest');
END;
/
---5.2 通过游标缓存从内存中读取sql填充
DECLARE
cur DBMS_SQLTUNE.SQLSET_CURSOR;
BEGIN
OPEN cur FOR
SELECT VALUE(x)
FROM table(
DBMS_SQLTUNE.SELECT_CURSOR_CACHE(
'parsing_schema_name <> ''SYS'' AND disk_reads > 1000000',
NULL, NULL, NULL, NULL, 1, NULL,'ALL')) x;
--
DBMS_SQLTUNE.LOAD_SQLSET(
sqlset_owner =>'SCOTT',
sqlset_name => 'OCPYANG_STS',
populate_cursor => cur);
END;
/
/**********常见错误
第 1 行出现错误:
ORA-13761: 过滤器无效
ORA-06512: 在 "SYS.DBMS_SQLTUNE", line 4715
ORA-06512: 在 line 11
使用SYS账户执行即可.
************************************/
SELECT sqlset_name, elapsed_time
,cpu_time, buffer_gets, disk_reads, sql_text
FROM dba_sqlset_statements
WHERE sqlset_name = 'OCPYANG_STS';
---5.3 将内存中指定时间内的所有sql加载
--语法:
DBMS_SQLTUNE.CAPTURE_CURSOR_CACHE_SQLSET (
sqlset_name IN VARCHAR2,
time_limit IN POSITIVE := 1800,
repeat_interval IN POSITIVE := 300,
capture_option IN VARCHAR2 := 'MERGE',
capture_mode IN NUMBER := MODE_REPLACE_OLD_STATS,
basic_filter IN VARCHAR2 := NULL,
sqlset_owner IN VARCHAR2 := NULL);
BEGIN
DBMS_SQLTUNE.CAPTURE_CURSOR_CACHE_SQLSET(
sqlset_owner =>'SCOTT'
, sqlset_name => 'PROD_WORKLOAD'
,time_limit => 3600 --3600秒
,repeat_interval => 20); --每隔20秒
END;
/
************************************************************
6.选择性的从sql调优集中删除sql
************************************************************
select sqlset_name, disk_reads, cpu_time, elapsed_time, buffer_gets
from dba_sqlset_statements;
BEGIN
DBMS_SQLTUNE.DELETE_SQLSET(
sqlset_owner => 'SCOTT',
sqlset_name => 'IO_STS'
,basic_filter => 'disk_reads < 2000000');
END;
/
************************************************************
7.传输sql调优集-STS
************************************************************
1.新建一个STS
---删除存在的STS
BEGIN
DBMS_SQLTUNE.DROP_SQLSET(
sqlset_name => 'OCPYANG_STS'
);
END;
/
---新建STS
BEGIN
DBMS_SQLTUNE.CREATE_SQLSET(
sqlset_name => 'OCPYANG_STS',
sqlset_owner => 'SCOTT',
description => 'ocpyangtest');
END;
/
---2.加载STS(可以参考STS收集的方法)
declare
baseline_ref_cur DBMS_SQLTUNE.SQLSET_CURSOR;
begin
open baseline_ref_cur for
select VALUE(p) from table(
DBMS_SQLTUNE.SELECT_WORKLOAD_REPOSITORY(
&begin_snap_id,
&end_snap_id,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
'ALL')) p;
DBMS_SQLTUNE.LOAD_SQLSET(
sqlset_name => 'my_sql_tuning_set',
populate_cursor => baseline_cursor);
end;
/
SELECT * FROM TABLE(DBMS_SQLTUNE.SELECT_SQLSET(
'my_sql_tuning_set',
'(disk_reads/buffer_gets) >= 0.75'));
---3.新建搜集baseline的表
BEGIN
dbms_spm.create_stgtab_baseline(
table_name => 'BASELINE_STG01',
table_owner => 'SCOTT',
db_version => DBMS_SQLTUNE.STS_STGTAB_11_2_VERSION
); --不能新建在SYS账户下
END;
/
----4.把Baseline数据填到表
BEGIN
DBMS_SQLTUNE.pack_stgtab_sqlset(
sqlset_name => 'OCPYANG_STS',
sqlset_owner => 'SCOTT',
staging_table_name => 'BASELINE_STG01',
staging_schema_owner => 'SCOTT',
db_version => DBMS_SQLTUNE.STS_STGTAB_11_2_VERSION );
END;
/
/**********语法
DBMS_SQLTUNE.PACK_STGTAB_SQLSET (
sqlset_name IN VARCHAR2,
sqlset_owner IN VARCHAR2 := NULL,
staging_table_name IN VARCHAR2,
staging_schema_owner IN VARCHAR2 := NULL,
db_version IN NUMBER := NULL);
***********/
---5.传递数据到目标服务器
使用Oracle Data Pump or database link or expdp等将表BASELINE_STG01
迁移到目标服务器.
---6.目标服务器新建STS
BEGIN
DBMS_SQLTUNE.CREATE_SQLSET(
sqlset_name => 'OCPYANG_STS01',
sqlset_owner => 'SCOTT',
description => 'ocpyangtest');
END;
/
----7.导入数据到目标服务器的STS
BEGIN
DBMS_SQLTUNE.UNPACK_STGTAB_SQLSET(
sqlset_name => 'OCPYANG_STS01',
replace => TRUE,
staging_table_name => 'BASELINE_STG01');
END;
/
---8.通过SPM BASELINE的包来把SQL调优集里的SQL都批量的生成BASELINE
declare
ret number;
begin
ret := dbms_spm.load_plans_from_sqlset(
sqlset_name => 'OCPYANG_STS01',
sqlset_owner => 'SCOTT');
end;
/
标签：sql调优集
原文地址：http://blog.csdn.net/yangzhawen/article/details/43736541
