DB2數據庫慢SQL性能優化詳解

一、抓取批量耗時最久的SQL：

2、執行次數最多的TOP10SQL"

db2 "select substr(stmt_text,1,500) as stmt_text,NUM_EXECUTIONS,TOTAL_USR_CPU_TIME,TOTAL_EXEC_TIME,SORT_OVERFLOWS from sysibmadm.snapdyn_sql order by NUM_EXECUTIONS desc fetch first 10 rows only"|grep -iv selected

3、平均執行時間最長的TOP10SQL"

db2 "select substr(stmt_text,1,500) as stmt_text,average_execution_time_s,num_executions from sysibmadm.top_dynamic_sql order by average_execution_time_s desc fetch first 10 rows only"|grep -iv selected

4、排序最多的TOP10SQL"

db2 "select substr(stmt_text,1,500) as stmt_text,stmt_sorts,SORT_OVERFLOWS,TOTAL_USR_CPU_TIME,NUM_EXECUTIONS from sysibmadm.snapdyn_sql order by stmt_sorts desc fetch first 10 rows only"|grep -iv selected

5、通過以上SQL獲得最耗時的SQL信息如下：

update tbhisfeyehz t set client_num=( select count(distinct a.in_client_no) from vsharedetail a, tbbranch b, tbtainfo c where a.ta_code=c.ta_code and a.open_branch=b.branch_no and substr(b.internal_branch,1,length(Rtrim(t.internal_branch))) = t.internal_branch and a.client_type=t.client_type and a.prd_code=t.prd_code and a.ta_code=t.ta_code and c.prd_type=t.prd_type and a.tot_vol > 0.0001 and a.ta_code=?) where exists(select 1 from vsharedetail a, tbbranch b, tbtainfo c where a.ta_code=c.ta_code and a.open_branch=b.branch_no and substr(b.internal_branch,1,length(Rtrim(t.internal_branch))) = t.internal_branch and a.client_type=t.client_type and a.prd_code=t.prd_code and a.ta_code=t.ta_code and c.prd_type=t.prd_type and a.tot_vol > 0.0001 and a.ta_code=?) and internal_branch in(select internal_branch from tbbranch where branch_level<>?) and sum_flag='0clear

三、爲了在測試環境復現該執行計劃我們需要從生產導出該SQL用到的所有基表的統計信息還原到測試環境下去。

1、導出統計信息：

db2look -d ifm30 -u ifm40 -m -t tbhisfeyehz -o tbhisfeyehz.sql

db2look -d ifm30 -u ifm40 -m -t tbsharedetail1 -o tbsharedetail1.sql

db2look -d ifm30 -u ifm40 -m -t tbsharedetail2 -o tbsharedetail2.sql

db2look -d ifm30 -u ifm40 -m -t tbsharedetail3 -o tbsharedetail3.sql

db2look -d ifm30 -u ifm40 -m -t tbsharedetail4 -o tbsharedetail4.sql

db2look -d ifm30 -u ifm40 -m -t tbsharedetail5 -o tbsharedetail5.sql

db2look -d ifm30 -u ifm40 -m -t tsys_branch -o tsys_branch.sql

db2look -d ifm30 -u ifm40 -m -t tbtainfo -o tbtainfo.sql

2、生效到測試環境：

db2 -tvf tbsharedetail1.sql

db2 -tvf tbsharedetail2.sql

db2 -tvf tbsharedetail3.sql

db2 -tvf tbsharedetail4.sql

db2 -tvf tbsharedetail5.sql

db2 -tvf tsys_branch.sql

db2 -tvf tbtainfo.sql

91pron free videos

db2 -tvf tbhisfeyehz.sql

四、分析SQL執行計劃，根據執行計劃分析關鍵執行成本之處的SQL計劃並做出調整計劃

db2 connect to ifm30

db2 set current explain mode explain

db2 -tvf optim.sql

db2 set current explain mode no

db2exfmt -d ifm30 -l -o

五、使用優化器顧問程序給出的建議進行優化

db2advis -d ifm30 -i optim3.sql -m i -l -1 -t 0 -o product.txt

六、優化相關基表

1、根據優化引擎建議新建唯一索引在對應基表之上

-- index[1], 0.013MB

CREATE INDEX "IFM40 "."IDX1501210621430" ON "IFM40 "."TBSHAREDETAIL4"