重庆分公司,新征程启航
为企业提供网站建设、域名注册、服务器等服务
性能分析过程中,经常会遇到生产库出现SQL的性能问题,但是,我们没有办法在生产库上做很多动作,需要将这个SQL的对应的表结构信息,统计信息导入到测试库进行测试(没有真实的测试数据,只有统计信息)
让客户满意是我们工作的目标,不断超越客户的期望值来自于我们对这个行业的热爱。我们立志把好的技术通过有效、简单的方式提供给客户,将通过不懈努力成为客户在信息化领域值得信任、有价值的长期合作伙伴,公司提供的服务项目有:主机域名、网络空间、营销软件、网站建设、兰山网站维护、网站推广。
本次测试目的,客户环境导出了Test和product环境的相同SQL的不同执行计划的SQLT报告。
TEST环境是11.2.0.1版本
Product是11.2.0.2版本
我们会将两个SQLT报告分别导入到11.2.0.1和11.2.0.2的测试环境中,然后将11.2.0.1版本,好的执行计划,作为profile导出到11.2.0.2中。让11.2.0.2也根据11.2.0.1中的SQL一样,使用相同的执行计划。
~~~~~~~第一部分,导入testcase环境,只导入统计信息,和表的元数据。没有真实数据~~~~~~
1. 参考sqlt_s39917_readme.html(客户SQLT输出的文件中,就已经包含这个readme.html文件)
也可以参考文档How to Use SQLT (SQLTXPLAIN) to Create a Testcase Without Row Data (Doc ID 1470811.1)
Implement SQLT Test Case (TC) Custom mode Create test case user and schema objects connecting as SYSDBA: sqlplus / as sysdba START sqlt_s39917_metadata.sql Purge pre-existing s39917 from local SQLT repository connected as SYSDBA: START sqlt_s39917_purge.sql Import SQLT repository for s39917 (provide SQLTXPLAIN password): HOS imp sqltxplain FILE=sqlt_s39917_exp.dmp LOG=sqlt_s39917_imp.log TABLES=sqlt% IGNORE=Y Restore CBO schema statistics for test case user connected as SYSDBA: START sqlt_s39917_restore.sql Restore CBO system statistics connected as SYSDBA: START sqlt_s39917_system_stats.sql Set the CBO environment connecting as test case user TC39917 (include optional test case user suffix): CONN TC39917/TC39917 START sqlt_s39917_set_cbo_env.sql Execute test case: START tc.sql
2. 执行tc.sql
注意:每次sqlplus连接进入后,都需要执行最后两步
CONN TC39917/TC39917 START sqlt_s39917_set_cbo_env.sql Execute test case: START tc.sql ----需要修改q.sql中对应表的用户名为TC39917
TC.sql执行结果如下
SQL> start tc.sql SQL> REM Executes SQL on TC then produces execution plan. Just execute "@tc.sql" from sqlplus. SQL> SET APPI OFF SERVEROUT OFF; SQL> @@q.sql SQL> SELECT Distinct fc_acc.blnctr_acc, 2 fc_acc.shortcut, 3 fc_acc.acc_name, 4 fc_subject.amount_money_sign, 5 fc_acc.subject_code, 6 fc_acc.corp_code, 7 fc_acc.net_code 8 FROM TC39917.fc_acc, TC39917.fc_user_corp, TC39917.fc_subject 9 WHERE ((fc_acc.corp_code = fc_user_corp.corp_code) OR 10 (acc_cussent is not Null or acc_cussent <> '')) 11 AND (fc_acc.subject_code = fc_subject.subject_code) 12 and ((fc_acc.checker_code is Not Null)) 13 AND (fc_acc.acc_state <> 'R') 14 AND (fc_acc.cur_code = 01) 15 and (fc_acc.acc_state <> 'D') 16 AND (fc_acc.net_code = 1000) 17 AND (fc_acc.acc_type_code = '02' or fc_subject.sys_subject_Code = '1010' or 18 (union_acc is not Null or union_acc <> '')) ; no rows selected SQL> @@plan.sql SQL> REM Displays plan for most recently executed SQL. Just execute "@plan.sql" from sqlplus. SQL> SET PAGES 2000 LIN 180; SQL> SPO plan.log; SQL> --SELECT * FROM TABLE(DBMS_XPLAN.DISPLAY_CURSOR); SQL> SELECT * FROM TABLE(DBMS_XPLAN.DISPLAY_CURSOR(NULL,NULL,'BASIC ROWS COST PREDICATE')); PLAN_TABLE_OUTPUT ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ EXPLAINED SQL STATEMENT: ------------------------ SELECT Distinct fc_acc.blnctr_acc, fc_acc.shortcut, fc_acc.acc_name, fc_subject.amount_money_sign, fc_acc.subject_code, fc_acc.corp_code, fc_acc.net_code FROM fc_acc, fc_user_corp, fc_subject WHERE ((fc_acc.corp_code = fc_user_corp.corp_code) OR (acc_cussent is not Null or acc_cussent <> '')) AND (fc_acc.subject_code = fc_subject.subject_code) and ((fc_acc.checker_code is Not Null)) AND (fc_acc.acc_state <> 'R') AND (fc_acc.cur_code = 01) and (fc_acc.acc_state <> 'D') AND (fc_acc.net_code = 1000) AND (fc_acc.acc_type_code = '02' or fc_subject.sys_subject_Code = '1010' or (union_acc is not Null or union_acc <> '')) Plan hash value: 31035937 ------------------------------------------------------------------------------ | Id | Operation | Name | Rows | Cost (%CPU)| ------------------------------------------------------------------------------ | 0 | SELECT STATEMENT | | | 133 (100)| | 1 | HASH UNIQUE | | 4573K| 133 (3)| | 2 | CONCATENATION | | | | | 3 | MERGE JOIN CARTESIAN | | 1 | 42 (0)| | 4 | NESTED LOOPS | | | | | 5 | NESTED LOOPS | | 1 | 36 (0)| |* 6 | TABLE ACCESS FULL | FC_ACC | 1 | 35 (0)| |* 7 | INDEX UNIQUE SCAN | PK_FC_SUBJECT | 1 | 0 (0)| |* 8 | TABLE ACCESS BY INDEX ROWID| FC_SUBJECT | 1 | 1 (0)| | 9 | BUFFER SORT | | 6210 | 41 (0)| | 10 | TABLE ACCESS FULL | FC_USER_CORP | 6210 | 6 (0)| | 11 | MERGE JOIN CARTESIAN | | 1 | 42 (0)| | 12 | NESTED LOOPS | | | | | 13 | NESTED LOOPS | | 1 | 36 (0)| |* 14 | TABLE ACCESS FULL | FC_ACC | 1 | 35 (0)| |* 15 | INDEX UNIQUE SCAN | PK_FC_SUBJECT | 1 | 0 (0)| |* 16 | TABLE ACCESS BY INDEX ROWID| FC_SUBJECT | 1 | 1 (0)| | 17 | BUFFER SORT | | 6210 | 41 (0)| | 18 | TABLE ACCESS FULL | FC_USER_CORP | 6210 | 6 (0)| |* 19 | HASH JOIN | | 122 | 47 (3)| |* 20 | HASH JOIN | | 7 | 41 (3)| |* 21 | TABLE ACCESS FULL | FC_ACC | 8 | 35 (0)| | 22 | TABLE ACCESS FULL | FC_SUBJECT | 636 | 5 (0)| | 23 | TABLE ACCESS FULL | FC_USER_CORP | 6210 | 6 (0)| ------------------------------------------------------------------------------ Predicate Information (identified by operation id): --------------------------------------------------- 6 - filter((TO_NUMBER("FC_ACC"."CUR_CODE")=1 AND "FC_ACC"."ACC_STATE"<>'R' AND "FC_ACC"."ACC_STATE"<>'D' AND "ACC_CUSSENT"<>'' AND "FC_ACC"."CHECKER_CODE" IS NOT NULL AND TO_NUMBER("FC_ACC"."NET_CODE")=1000)) 7 - access("FC_ACC"."SUBJECT_CODE"="FC_SUBJECT"."SUBJECT_CODE") 8 - filter(("FC_ACC"."ACC_TYPE_CODE"='02' OR "FC_SUBJECT"."SYS_SUBJECT_CODE"='1010' OR "UNION_ACC" IS NOT NULL OR "UNION_ACC"<>'')) 14 - filter(("ACC_CUSSENT" IS NOT NULL AND TO_NUMBER("FC_ACC"."CUR_CODE")=1 AND "FC_ACC"."ACC_STATE"<>'R' AND "FC_ACC"."ACC_STATE"<>'D' AND "FC_ACC"."CHECKER_CODE" IS NOT NULL AND TO_NUMBER("FC_ACC"."NET_CODE")=1000 AND LNNVL("ACC_CUSSENT"<>''))) 15 - access("FC_ACC"."SUBJECT_CODE"="FC_SUBJECT"."SUBJECT_CODE") 16 - filter(("FC_ACC"."ACC_TYPE_CODE"='02' OR "FC_SUBJECT"."SYS_SUBJECT_CODE"='1010' OR "UNION_ACC" IS NOT NULL OR "UNION_ACC"<>'')) 19 - access("FC_ACC"."CORP_CODE"="FC_USER_CORP"."CORP_CODE") 20 - access("FC_ACC"."SUBJECT_CODE"="FC_SUBJECT"."SUBJECT_CODE") filter(("FC_ACC"."ACC_TYPE_CODE"='02' OR "FC_SUBJECT"."SYS_SUBJECT_CODE"='1010' OR "UNION_ACC" IS NOT NULL OR "UNION_ACC"<>'')) 21 - filter((TO_NUMBER("FC_ACC"."CUR_CODE")=1 AND "FC_ACC"."ACC_STATE"<>'R' AND "FC_ACC"."ACC_STATE"<>'D' AND "FC_ACC"."CHECKER_CODE" IS NOT NULL AND TO_NUMBER("FC_ACC"."NET_CODE")=1000 AND LNNVL("ACC_CUSSENT" IS NOT NULL) AND LNNVL("ACC_CUSSENT"<>''))) 76 rows selected. SQL> SPO OFF; SQL>
假设,上面的执行计划是我们期望的,性能优异的。我需要导出为profile,并导入到性能较差的数据库
~~~~~~~~第二部分,导出11.2.0.1的profile~~~~~~~~~~~~~~~~~~~~
3. 生成profile
SQL> @/home/oracle/Bright-temp/sqlt/utl/coe_xfr_sql_profile.sql cwju3syt2mx9a 31035937 SQL> SPO coe_xfr_sql_profile.log; SQL> SET DEF ON TERM OFF ECHO ON FEED OFF VER OFF HEA ON LIN 2000 PAGES 100 LONG 8000000 LONGC 800000 TRIMS ON TI OFF TIMI OFF SERVEROUT ON SIZE 1000000 NUMF "" SQLP SQL>; Parameter 1: SQL_ID (required) PLAN_HASH_VALUE AVG_ET_SECS --------------- ----------- 31035937 .014 Parameter 2: PLAN_HASH_VALUE (required) Values passed to coe_xfr_sql_profile: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ SQL_ID : "cwju3syt2mx9a" PLAN_HASH_VALUE: "31035937"
检查生成的profile执行文件,需要修改force_match => TRUE ,并且在新的数据库执行前,还需要针对新的数据库的testcase用户名,修改其中对应的表的用户名。
vi coe_xfr_sql_profile_cwju3syt2mx9a_31035937.sql
~~~~~~~~第三部分,导入性能差的SQLT到11.2.0.2,并查看执行计划~~~~~~~~~~~~~~
4. 导入SQLT_product到11.2.0.2上,查看11.2.0.2上这个SQL的执行计划
SQL> SPO OFF; SQL> START tc.sql SQL> REM Executes SQL on TC then produces execution plan. Just execute "@tc.sql" from sqlplus. SQL> SET APPI OFF SERVEROUT OFF; SQL> @@q.sql SQL> SELECT Distinct fc_acc.blnctr_acc, 2 fc_acc.shortcut, 3 fc_acc.acc_name, 4 fc_subject.amount_money_sign, 5 fc_acc.subject_code, 6 fc_acc.corp_code, 7 fc_acc.net_code 8 FROM TC34646.fc_acc, TC34646.fc_user_corp, TC34646.fc_subject 9 WHERE ((fc_acc.corp_code = fc_user_corp.corp_code) OR 10 (acc_cussent is not Null or acc_cussent <> '')) 11 AND (fc_acc.subject_code = fc_subject.subject_code) 12 and ((fc_acc.checker_code is Not Null)) 13 AND (fc_acc.acc_state <> 'R') 14 AND (fc_acc.cur_code = 01) 15 and (fc_acc.acc_state <> 'D') 16 AND (fc_acc.net_code = 1000) 17 AND (fc_acc.acc_type_code = '02' or fc_subject.sys_subject_Code = '1010' or 18 (union_acc is not Null or union_acc <> '')); no rows selected SQL> @@plan.sql SQL> REM Displays plan for most recently executed SQL. Just execute "@plan.sql" from sqlplus. SQL> SET PAGES 2000 LIN 180; SQL> SPO plan.log; SQL> --SELECT * FROM TABLE(DBMS_XPLAN.DISPLAY_CURSOR); SQL> SELECT * FROM TABLE(DBMS_XPLAN.DISPLAY_CURSOR(NULL,NULL,'BASIC ROWS COST PREDICATE')); PLAN_TABLE_OUTPUT ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ EXPLAINED SQL STATEMENT: ------------------------ SELECT Distinct fc_acc.blnctr_acc, fc_acc.shortcut, fc_acc.acc_name, fc_subject.amount_money_sign, fc_acc.subject_code, fc_acc.corp_code, fc_acc.net_code FROM TC34646.fc_acc, TC34646.fc_user_corp, TC34646.fc_subject WHERE ((fc_acc.corp_code = fc_user_corp.corp_code) OR (acc_cussent is not Null or acc_cussent <> '')) AND (fc_acc.subject_code = fc_subject.subject_code) and ((fc_acc.checker_code is Not Null)) AND (fc_acc.acc_state <> 'R') AND (fc_acc.cur_code = 01) and (fc_acc.acc_state <> 'D') AND (fc_acc.net_code = 1000) AND (fc_acc.acc_type_code = '02' or fc_subject.sys_subject_Code = '1010' or (union_acc is not Null or union_acc <> '')) Plan hash value: 561502789 -------------------------------------------------------------------- | Id | Operation | Name | Rows | Cost (%CPU)| -------------------------------------------------------------------- | 0 | SELECT STATEMENT | | | 278K(100)| | 1 | HASH UNIQUE | | 7469K| 278K (1)| |* 2 | HASH JOIN | | 7469K| 3321 (6)| |* 3 | TABLE ACCESS FULL | FC_ACC | 1156 | 38 (3)| | 4 | MERGE JOIN CARTESIAN| | 4189K| 3197 (3)| | 5 | TABLE ACCESS FULL | FC_SUBJECT | 636 | 5 (0)| | 6 | BUFFER SORT | | 6587 | 3192 (3)| | 7 | TABLE ACCESS FULL | FC_USER_CORP | 6587 | 5 (0)| -------------------------------------------------------------------- Predicate Information (identified by operation id): --------------------------------------------------- 2 - access("FC_ACC"."SUBJECT_CODE"="FC_SUBJECT"."SUBJECT_CODE") filter((("FC_ACC"."CORP_CODE"="FC_USER_CORP"."CORP_CODE" OR "ACC_CUSSENT" IS NOT NULL OR "ACC_CUSSENT"<>'') AND ("FC_ACC"."ACC_TYPE_CODE"='02' OR "FC_SUBJECT"."SYS_SUBJECT_CODE"='1010' OR "UNION_ACC" IS NOT NULL OR "UNION_ACC"<>''))) 3 - filter((TO_NUMBER("FC_ACC"."CUR_CODE")=1 AND "FC_ACC"."ACC_STATE"<>'D' AND "FC_ACC"."CHECKER_CODE" IS NOT NULL AND "FC_ACC"."ACC_STATE"<>'R' AND TO_NUMBER("FC_ACC"."NET_CODE")=1000)) 42 rows selected. SQL> SPO OFF;
5. 在将11.2.0.1的profile导入到11.2.0.2,然后查看profile是否生效
执行coe_xfr_sql_profile_cwju3syt2mx9a_31035937.sql前,选需要修改SQL文本为
SELECT Distinct fc_acc.blnctr_acc, fc_acc.shortcut, fc_acc.acc_name, fc_subject.amount_money_sign, fc_acc.subject_code, fc_acc.corp_code, fc_acc.net_code FROM TC34646.fc_acc, <<<<<<<<<<<<<<<<<<<<增加用户TC34646 TC34646.fc_user_corp, TC34646.fc_subject
6. 执行coe脚本
sqlplus / as sysdba SQL>@/home/oracle/Bright-temp/sqlt_s39917_xecute_TEST/coe_xfr_sql_profile_cwju3syt2mx9a_31035937.sql
7. 再次执行tc.sql
SQL> CONN TC34646/TC34646 SQL> START sqlt_s34646_set_cbo_env.sql SQL> START tc.sql SQL> REM Executes SQL on TC then produces execution plan. Just execute "@tc.sql" from sqlplus. SQL> SET APPI OFF SERVEROUT OFF; SQL> @@q.sql SQL> SELECT Distinct fc_acc.blnctr_acc, 2 fc_acc.shortcut, 3 fc_acc.acc_name, 4 fc_subject.amount_money_sign, 5 fc_acc.subject_code, 6 fc_acc.corp_code, 7 fc_acc.net_code 8 FROM TC34646.fc_acc, TC34646.fc_user_corp, TC34646.fc_subject 9 WHERE ((fc_acc.corp_code = fc_user_corp.corp_code) OR 10 (acc_cussent is not Null or acc_cussent <> '')) 11 AND (fc_acc.subject_code = fc_subject.subject_code) 12 and ((fc_acc.checker_code is Not Null)) 13 AND (fc_acc.acc_state <> 'R') 14 AND (fc_acc.cur_code = 01) 15 and (fc_acc.acc_state <> 'D') 16 AND (fc_acc.net_code = 1000) 17 AND (fc_acc.acc_type_code = '02' or fc_subject.sys_subject_Code = '1010' or 18 (union_acc is not Null or union_acc <> '')); no rows selected SQL> @@plan.sql SQL> REM Displays plan for most recently executed SQL. Just execute "@plan.sql" from sqlplus. SQL> SET PAGES 2000 LIN 180; SQL> SPO plan.log; SQL> --SELECT * FROM TABLE(DBMS_XPLAN.DISPLAY_CURSOR); SQL> SELECT * FROM TABLE(DBMS_XPLAN.DISPLAY_CURSOR(NULL,NULL,'BASIC ROWS COST PREDICATE')); PLAN_TABLE_OUTPUT ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ EXPLAINED SQL STATEMENT: ------------------------ SELECT Distinct fc_acc.blnctr_acc, fc_acc.shortcut, fc_acc.acc_name, fc_subject.amount_money_sign, fc_acc.subject_code, fc_acc.corp_code, fc_acc.net_code FROM TC34646.fc_acc, TC34646.fc_user_corp, TC34646.fc_subject WHERE ((fc_acc.corp_code = fc_user_corp.corp_code) OR (acc_cussent is not Null or acc_cussent <> '')) AND (fc_acc.subject_code = fc_subject.subject_code) and ((fc_acc.checker_code is Not Null)) AND (fc_acc.acc_state <> 'R') AND (fc_acc.cur_code = 01) and (fc_acc.acc_state <> 'D') AND (fc_acc.net_code = 1000) AND (fc_acc.acc_type_code = '02' or fc_subject.sys_subject_Code = '1010' or (union_acc is not Null or union_acc <> '')) Plan hash value: 31035937 ------------------------------------------------------------------------------ | Id | Operation | Name | Rows | Cost (%CPU)| ------------------------------------------------------------------------------ | 0 | SELECT STATEMENT | | | 8755 (100)| | 1 | HASH UNIQUE | | 7469K| 8755 (1)| | 2 | CONCATENATION | | | | | 3 | MERGE JOIN CARTESIAN | | 226K| 245 (3)| | 4 | NESTED LOOPS | | | | | 5 | NESTED LOOPS | | 34 | 73 (2)| |* 6 | TABLE ACCESS FULL | FC_ACC | 35 | 38 (3)| |* 7 | INDEX UNIQUE SCAN | PK_FC_SUBJECT | 1 | 0 (0)| |* 8 | TABLE ACCESS BY INDEX ROWID| FC_SUBJECT | 1 | 1 (0)| | 9 | BUFFER SORT | | 6587 | 244 (3)| | 10 | TABLE ACCESS FULL | FC_USER_CORP | 6587 | 5 (0)| | 11 | MERGE JOIN CARTESIAN | | 2285 | 46 (3)| | 12 | NESTED LOOPS | | | | | 13 | NESTED LOOPS | | 1 | 39 (3)| |* 14 | TABLE ACCESS FULL | FC_ACC | 1 | 38 (3)| |* 15 | INDEX UNIQUE SCAN | PK_FC_SUBJECT | 1 | 0 (0)| |* 16 | TABLE ACCESS BY INDEX ROWID| FC_SUBJECT | 1 | 1 (0)| | 17 | BUFFER SORT | | 6587 | 45 (3)| | 18 | TABLE ACCESS FULL | FC_USER_CORP | 6587 | 7 (0)| |* 19 | HASH JOIN | | 193 | 51 (4)| |* 20 | HASH JOIN | | 11 | 43 (3)| |* 21 | TABLE ACCESS FULL | FC_ACC | 11 | 38 (3)| | 22 | TABLE ACCESS FULL | FC_SUBJECT | 636 | 5 (0)| | 23 | TABLE ACCESS FULL | FC_USER_CORP | 6587 | 7 (0)| ------------------------------------------------------------------------------ Predicate Information (identified by operation id): --------------------------------------------------- 6 - filter((TO_NUMBER("FC_ACC"."CUR_CODE")=1 AND "ACC_CUSSENT"<>'' AND "FC_ACC"."ACC_STATE"<>'D' AND "FC_ACC"."CHECKER_CODE" IS NOT NULL AND "FC_ACC"."ACC_STATE"<>'R' AND TO_NUMBER("FC_ACC"."NET_CODE")=1000)) 7 - access("FC_ACC"."SUBJECT_CODE"="FC_SUBJECT"."SUBJECT_CODE") 8 - filter(("FC_ACC"."ACC_TYPE_CODE"='02' OR "FC_SUBJECT"."SYS_SUBJECT_CODE"='1010' OR "UNION_ACC" IS NOT NULL OR "UNION_ACC"<>'')) 14 - filter(("ACC_CUSSENT" IS NOT NULL AND TO_NUMBER("FC_ACC"."CUR_CODE")=1 AND "FC_ACC"."ACC_STATE"<>'D' AND "FC_ACC"."CHECKER_CODE" IS NOT NULL AND "FC_ACC"."ACC_STATE"<>'R' AND TO_NUMBER("FC_ACC"."NET_CODE")=1000 AND LNNVL("ACC_CUSSENT"<>''))) 15 - access("FC_ACC"."SUBJECT_CODE"="FC_SUBJECT"."SUBJECT_CODE") 16 - filter(("FC_ACC"."ACC_TYPE_CODE"='02' OR "FC_SUBJECT"."SYS_SUBJECT_CODE"='1010' OR "UNION_ACC" IS NOT NULL OR "UNION_ACC"<>'')) 19 - access("FC_ACC"."CORP_CODE"="FC_USER_CORP"."CORP_CODE") 20 - access("FC_ACC"."SUBJECT_CODE"="FC_SUBJECT"."SUBJECT_CODE") filter(("FC_ACC"."ACC_TYPE_CODE"='02' OR "FC_SUBJECT"."SYS_SUBJECT_CODE"='1010' OR "UNION_ACC" IS NOT NULL OR "UNION_ACC"<>'')) 21 - filter((TO_NUMBER("FC_ACC"."CUR_CODE")=1 AND "FC_ACC"."ACC_STATE"<>'D' AND "FC_ACC"."CHECKER_CODE" IS NOT NULL AND "FC_ACC"."ACC_STATE"<>'R' AND TO_NUMBER("FC_ACC"."NET_CODE")=1000 AND LNNVL("ACC_CUSSENT" IS NOT NULL) AND LNNVL("ACC_CUSSENT"<>''))) 74 rows selected. SQL> SPO OFF; SQL>
最终,可以看到,在11.2.0.2上,也按照11.2.0.1上的执行计划进行执行SQL了。
这里要强调一下SQLT真的是非常使用的SQL 调优工具。而且非常易用,很多步骤都在SQLT的HTML文件中,逐步指导。建议大家学习使用。
参考文档:
FAQ: SQLT (SQLTXPLAIN) Frequently Asked Questions (Doc ID 1454160.1)
All About the SQLT Diagnostic Tool (Doc ID 215187.1)
How to Collect Standard Diagnostic Information Using SQLT for SQL Issues (Doc ID 1683772.1)
SQLT Usage Instructions (Doc ID 1614107.1)
How to Generate a SQLT of a SQL Statement That Hangs or Raises an Error (Doc ID 2146816.1)
How to Use SQLT (SQLTXPLAIN) to Create a Testcase Without Row Data (Doc ID 1470811.1)