当前位置: 首页 > 图文教程 > 数据库 > Oracle > Performance Improvement Tips for Oracle on UNIX

Oracle
Oracle数据库技术(32)
Oracle数据库技术(33)
Oracle数据库技术(34)
Oracle数据库技术(35)
Oracle数据库技术(36)
Oracle数据安全面面观
Oracle数据操作和控制语言详解
Oracle数据库数据对象分析
解析Oracle 8i/9i的计划稳定性
使用Oracle实现实时通信
Oracle数据库中索引的维护
Oracle数据库游标使用大全
Oracle9i中监视索引的使用
在Oracle9i中使用多种Block Size
监控Oracle数据库的常用shell脚本
Performance Improvement Tips for Oracle on UNIX
Raw Partitions and Windows NT
How to use OS commands to diagnose Database Performance issues?
Raw Devices and Oracle - 20 Common Questions and Answers
Monitor Oracle Resource Consumption in UNIX

Performance Improvement Tips for Oracle on UNIX


出处:互联网   整理: 软晨网(RuanChen.com)   发布: 2009-10-31   浏览: 35 ::
收藏到网摘: n/a

TIMED_STATISTICS = true

  1. Restart your database. 
  2. Run your application.
  3. Run tkprof against the tracefile created by your application:

tkprof EXPLAIN=username/passwd

  1. Look at formatted output of trace command and make sure that your

      SQL statement is using indexes correctly.  Refer to DBA guide for a list

      of rules that the oracle optimizer uses when choosing a path for a SQL statement. 

                 <<<<<< OUTPUT OF TKPROF FILE >>>>>>> 

count   = number of times OPI procedure was executed 

cpu     = cpu time executing in hundredths of seconds 

elap    = elapsed time executing in hundredths of secs 

phys    = number of physical reads of buffers (from disk) 

cr      = number of buffers gotten for consistent read 

cur     = number of buffers gotten in current mode (usually for update) 

rows    = number of rows processed by the OPI call 

===========================================================

select * from emp where empno=7369  

            count     cpu    elap    phys      cr     cur    rows 

Parse:          1       0       0       0       0       0 

Execute:        1       0       0       0       0       2       0 

Fetch:          1       0       0     219     227       0       2 

Execution plan: 

TABLE ACCESS (FULL) OF EMP 

===========================================================

select empno from emp where empno=7934  

            count     cpu    elap    phys      cr     cur    rows 

Parse:          2       0       0       0       0       0 

Execute:        2       0       0       0       0       2       0 

Fetch:          2       0       0       0&nb