这篇文章主要介绍“Oracle SQL执行计划异常的处理方法”,在日常操作中,相信很多人在Oracle SQL执行计划异常的处理方法问题上存在疑惑,小编查阅了各式资料,整理出简单好用的操作方法,希望对大家解答”Oracle SQL执行计划异常的处理方法”的疑惑有所帮助!接下来,请跟着小编一起来学习吧!
现象:
下面语句一直以来都比较高效,执行计划用了索引范围扫描后经历三次嵌套循环,可在2秒内返回结果,但今天经同事反映却走了1分多钟!
原SQL语句:
Select * From (Select Rownum As Rownumber__, t.* From (Select T1.Orderdate As "OrderDate", T1.Status As "Status", T1.Ordercode As "OrderCode", T1.Sumamt As "SumAmt", T1.Ordertype As "OrderType", T1.Questiondesc As "QuestionDesc", T1.Ordersource As "OrderSource", T2.Accepter As "Accepter", T2.City As "City", T1.Isquestion As "IsQuestion", T1.Issplit As "IsSplit", T1.Salemode As "SaleMode", T1.Stockout As "StockOut", T2.Encmobile As "EncMobile", T2.Encphone As "EncPhone", Decryptbykey(T2.Mobilephone) As "MobilePhone", T2.Province As "Province", T3.Checkercode As "CheckerCode", T3.Iscancel As "IsCancel", T3.Ischeck As "IsCheck", T3.Isclose As "IsClose", T3.Isfinish As "IsFinish", T1.Ischange As "IsChange" From Xs_Order T1 Join Xs_Orderpsaddress T2 On T1.Ordercode = T2.Ordercode Join Xs_Orderstatus T3 On T1.Ordercode = T3.Ordercode Order By T1.Ordercode Desc) t Where "OrderDate" >= :Orderdate0 And "StockOut" = :Stockout1) Temp Where Rownumber__ > 0and Rownumber__ <= 20
后来查看执行计划,执行计划变成:
分析:
由于后面两个表是大表,全表扫描导致大量的IO消耗,该语句采用了绑定变量,如果把绑定变量调整为常量后,执行计划正常走了索引连接,执行后返回也是在2秒内。曾经以为是绑定变量窥探异常问题,后来把表的统计信息重新更新后,问题依旧,接着运行SQL TUNNING包,概要只建议说要启用并行,但全表扫描并没有消除,考虑到代价太高就放弃,于是想用DBMS_SPM包来载入该语句:
BASELINE: declare l_pls number; begin l_pls := DBMS_SPM.LOAD_PLANS_FROM_CURSOR_CACHE( sql_id=> 'a0wawpy5hfrt3', plan_hash_value => 2253704843, --注意这里的2253704843是我用常量带入后正常的PLAN_HASH_VALUE enabled => 'YES'); end;
执行后,发现语句还是走了错误的执行计划,曾经考虑想用捕捉基线的方式进行演化,但由于该语句带绑定变量,会话级比较难搞,所以想到用包删除共享池里的该执行计划,让它重新进行硬解析:操作如下:
exec dbms_shared_pool.purge('0000000DE5E6B808,2332516131', 'c')
–第一个参数为v$sqlarea中address和hash_value,第二个为cursor类型)
处理后执行计划重新产生,并自动应用了2253704843这个执行计划,查询效率正常:
到此,关于“Oracle SQL执行计划异常的处理方法”的学习就结束了,希望能够解决大家的疑惑。理论与实践的搭配能更好的帮助大家学习,快去试试吧!若想继续学习更多相关知识,请继续关注亿速云网站,小编会继续努力为大家带来更多实用的文章!
免责声明:本站发布的内容(图片、视频和文字)以原创、转载和分享为主,文章观点不代表本网站立场,如果涉及侵权请联系站长邮箱:is@yisu.com进行举报,并提供相关证据,一经查实,将立刻删除涉嫌侵权内容。