如何获取真实的执行计划 （下）

续 如何获取真实的执行计划（上） http://bfc99.blog.51cto.com/265386/1706835

   方法4： 10046 trace跟踪

SQL> alter session set events '10046 trace name context  forever,level 12';  --打开跟踪

 

Session altered.

 

SQL> SELECT  * FROM t1, t2 WHERE t1.id = t2.t1_id AND t1.n in(18,19);  --执行相应的SQL

 

         ID         N

---------- ----------

CONTENTS

---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

-------------------------------------------------------------------------------------------------------------------------------------------------------------------

         ID     T1_ID              N

---------- ---------- ----------

CONTENTS

---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

-------------------------------------------------------------------------------------------------------------------------------------------------------------------

         18        18

yFfYpcNtsxEpvWUOceJmHtvpCEbcJTUTeKExdHlGlwoIFEgmOo

         18        18             18

OWCPIKKFQTMBZAVJBUGPOGOZGPHCMTGFDJTMCIZXRFVXYCATTY

 

         19        19

jAOYAJHBzwhNWHImZeFUOaxvMycbQsCLKhsUnSFwZpyctEahjK

         19        19             19

BEQUUMJSUDRLTCGOCIUHSZCNNJOTZOCEQKBZPSLHGAKRTLJAHX

 

 

SQL> alter session set events '10046 trace name context off';   --关闭跟踪。

 

Session altered.

 
--通过以下SQL，获取跟踪期间产生的跟踪文件的所在位置。

SQL> select d.value

|| '/'

|| LOWER (RTRIM(i.INSTANCE, CHR(0)))

|| '_ora_'

|| p.spid

|| '.trc' trace_file_name

from (select p.spid

      from v$mystat m,v$session s, v$process p

      where  m.statistic#=1 and s.sid=m.sid and p.addr=s.paddr) p,

  2    3    4    5    6    7    8    9   10        (select t.INSTANCE

       FROM v$thread t,v$parameter v

       WHERE v.name='thread'

       AND(v.VALUE=0 OR t.thread#=to_number(v.value))) i,

       (select value

       from v$parameter

       where name='user_dump_dest') d; 11   12   13   14   15   16 

 

TRACE_FILE_NAME

---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

-------------------------------------------------------------------------------------------------------------------------------------------------------------------

/oracle/app/oracle/admin/orcl/udump/orcl_ora_2509.trc

 

SQL> exit

Disconnected from Oracle Database 10g Enterprise Edition Release 10.2.0.5.0 - 64bit Production

With the Partitioning, OLAP, Data Mining and Real Application Testing options

--通过以下命令，进行跟踪文件的转换，以方便阅读。

[oracle@localhost ~]$ tkprof /oracle/app/oracle/admin/orcl/udump/orcl_ora_2509.trc 10046.txt sys=no sort=prsela,exeela,fchela

 

TKPROF: Release 10.2.0.5.0 - Production on Sun Apr 27 11:06:18 2014

 

Copyright (c) 1982, 2007, Oracle.  All rights reserved.

 

 

[oracle@localhost ~]$ ll

total 280

-rw-r--r-- 1 oracle oinstall   5537 Apr 27 11:06 10046.txt

-rw-r--r-- 1 oracle oinstall    797 Feb  9 12:14 demo.txt

-rw-r--r-- 1 oracle oinstall  30335 Nov 23 14:29 zyash.html

-rw-r--r-- 1 oracle oinstall 234760 Nov 23 14:04 zyawr.rpt

[oracle@localhost ~]$ cat 10046.txt

 

TKPROF: Release 10.2.0.5.0 - Production on Sun Apr 27 11:06:18 2014

 

Copyright (c) 1982, 2007, Oracle.  All rights reserved.

 

Trace file: /oracle/app/oracle/admin/orcl/udump/orcl_ora_2509.trc

Sort options: prsela  exeela  fchela 

********************************************************************************

count    = number of times OCI procedure was executed

cpu      = cpu time in seconds executing

elapsed  = elapsed time in seconds executing

disk     = number of physical reads of buffers from disk

query    = number of buffers gotten for consistent read

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

rows     = number of rows processed by the fetch or execute call

********************************************************************************

 

SELECT  *

FROM

 t1, t2 WHERE t1.id = t2.t1_id AND t1.n in(18,19)

 

 

call     count       cpu    elapsed       disk      query    current        rows

------- ------  -------- ---------- ---------- ---------- ----------  ----------

Parse        1      0.00       0.00          0          0          0           0

Execute      1      0.00       0.00          0          0          0           0

Fetch        2      0.00       0.00          0         14          0           2

------- ------  -------- ---------- ---------- ---------- ----------  ----------

total        4      0.00       0.00          0         14          0           2

 

Misses in library cache during parse: 0

Optimizer mode: ALL_ROWS

Parsing user id: 54 

 

Rows     Row Source Operation

-------  ---------------------------------------------------

      2  TABLE ACCESS BY INDEX ROWID T2 (cr=14 pr=0 pw=0 time=142 us)

      5   NESTED LOOPS  (cr=12 pr=0 pw=0 time=79 us)

      2    INLIST ITERATOR  (cr=7 pr=0 pw=0 time=53 us)

      2     TABLE ACCESS BY INDEX ROWID T1 (cr=7 pr=0 pw=0 time=46 us)

      2      INDEX RANGE SCAN T1_N (cr=5 pr=0 pw=0 time=33 us)(object id 51619)

      2    INDEX RANGE SCAN T2_T1_ID (cr=5 pr=0 pw=0 time=25 us)(object id 51620)

 

 

Elapsed times include waiting on following events:

  Event waited on                             Times   Max. Wait  Total Waited

  ----------------------------------------   Waited  ----------  ------------

  SQL*Net message to client                       2        0.00          0.00

  SQL*Net message from client                     2       30.60         30.60

 

特点：1、SQL语句只有被执行过，才能得到其执行计划。故，得到执行计划时间的快慢，取决于SQL语句运行时间的长短。有SQL语句执行结果的输出。

        2、获取执行计划的步骤较复杂。

        3、解析时间和执行时间分别列出

        4、可以看到相关的统计信息（逻辑读、物理读、写等，执行计划中的每一步执行时间精确到微秒）

        5、如果SQL语句中有函数调用，SQL中有SQL，将会都被列出

        6、看不到谓词信息

 

    方法5： awrsqrpt.sql
该方法主要用于获取已经被排出SHARED POOL，即，无法在v$sql中查询出来，但仍在AWR快照中的历史SQL。

SQL> @?/rdbms/admin/awrsqrpt.sql

 

Current Instance

~~~~~~~~~~~~~~~~

 

   DB Id    DB Name  Inst Num Instance

----------- ------------ -------- ------------

 1347935162 ORCL          1 orcl

 

 

Specify the Report Type

~~~~~~~~~~~~~~~~~~~~~~~

Would you like an HTML report, or a plain text report?

Enter 'html' for an HTML report, or 'text' for plain text

Defaults to 'html'

Enter value for report_type:

 

Type Specified:  html

 

 

Instances in this Workload Repository schema

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

 

   DB Id     Inst Num DB Name            Instance     Host

------------ -------- ------------ ------------ ------------

* 1347935162     1 ORCL        orcl   localhost.lo

                                                        caldomain

 

Using 1347935162 for database Id

Using               1 for instance number

 

 

Specify the number of days of snapshots to choose from

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Entering the number of days (n) will result in the most recent

(n) days of snapshots being listed.  Pressing without

specifying a number lists all completed snapshots.

 

 

Enter value for num_days: 1

 

Listing the last day's Completed Snapshots

 

                                                                 Snap

Instance     DB Name     Snap Id    Snap Started    Level

------------ ------------ --------- ------------------ -----

orcl       ORCL               533 27 Apr 2014 09:59       1

                                     534 27 Apr 2014 11:00       1

 

 

 

Specify the Begin and End Snapshot Ids

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Enter value for begin_snap: 533

Begin Snapshot Id specified: 533

 

Enter value for end_snap: 534

End   Snapshot Id specified: 534

 

 

 

 

Specify the SQL Id

~~~~~~~~~~~~~~~~~~

Enter value for sql_id: cshvm6ngravw0

SQL ID specified:  cshvm6ngravw0

 

Specify the Report Name

~~~~~~~~~~~~~~~~~~~~~~~

The default report file name is awrsqlrpt_1_533_534.html.  To use this name,

press to continue, otherwise enter an alternative.

 

Enter value for report_name:

 

Using the report name awrsqlrpt_1_533_534.html

 
--以下为生成的报告的内容：

 

WORKLOAD REPOSITORY SQL Report

Snapshot Period Summary

DB Name	DB Id	Instance	Inst num	Release	RAC	Host
ORCL	1347935162	orcl	1	10.2.0.5.0	NO	localhost.localdomain

 

	Snap Id	Snap Time	Sessions	Cursors/Session
Begin Snap:	533	27-Apr-14  09:59:56	22	2.1
End Snap:	534	27-Apr-14  11:00:57	22	4.0
Elapsed:		61.02 (mins)
DB Time:		0.10 (mins)

SQL Summary

SQL Id	Elapsed Time (ms)	Module	Action	SQL Text
cshvm6ngravw0	68	SQL*Plus		SELECT * FROM t1, t2 WHERE t1.id =  t2.t1_id AND t1.n in(18, 19)...

Back to Top

SQL ID: cshvm6ngravw0

• 1st Capture and Last Capture Snap IDs refer to     Snapshot IDs witin the snapshot range

• SELECT * FROM t1, t2 WHERE t1.id =     t2.t1_id AND t1.n in(18,19)

#	Plan Hash Value	Total Elapsed  Time(ms)	Executions	1st Capture Snap  ID	Last Capture  Snap ID
1	128660979	68	1	534	534

Back to Top

Plan 1(PHV: 128660979)

• Plan Statistics

• Execution Plan

Back to Top

Plan Statistics

• % Total DB Time is the Elapsed Time of the SQL     statement divided into the Total Database Time multiplied by 100

Stat Name	Statement Total	Per Execution	% Snap Total
Elapsed Time  (ms)	68	67.82	1.10
CPU Time (ms)	45	44.99	1.36
Executions	1
Buffer Gets	225	225.00	0.10
Disk Reads	0	0.00	0.00
Parse Calls	1	1.00	0.01
Rows	2	2.00
User I/O Wait  Time (ms)	0
Cluster Wait  Time (ms)	0
Application Wait  Time (ms)	0
Concurrency Wait  Time (ms)	0
Invalidations	0
Version Count	1
Sharable Mem(KB)	27

Back to Plan 1(PHV: 128660979) 
Back to Top

Execution Plan

Id	Operation	Name	Rows	Bytes	Cost (%CPU)	Time
0	SELECT STATEMENT				8 (100)
1	TABLE ACCESS BY INDEX ROWID	T2	1	2041	2 (0)	00:00:01
2	NESTED LOOPS		2	4198	8 (0)	00:00:01
3	INLIST ITERATOR
4	TABLE ACCESS BY INDEX ROWID	T1	2	116	4 (0)	00:00:01
5	INDEX RANGE SCAN	T1_N	2		2 (0)	00:00:01
6	INDEX RANGE SCAN	T2_T1_ID	1		1 (0)	00:00:01

• dynamic sampling used for this statement

Back to Plan 1(PHV: 128660979) 
Back to Top

Full SQL Text

SQL Id	SQL Text
cshvm6ngravw0	SELECT * FROM  t1, t2 WHERE t1.id = t2.t1_id AND t1.n in(18, 19)

Back to Top

 

特点：1、SQL语句只有被执行过，才能得到其执行计划。

      2、可以看到已经从SHARED_POOL中被清除的执行计划。

      3、看不到谓词信息。

综上，对各种获取执行计划的方法适用场景归纳如下：

1.  如果某SQL执行时间较长，而又需要尽快看到执行计划，可以用explain plan for和dbms_xplan.display_cursor指定SQL_ID的方法；

       2. 跟踪某条SQL最简单的方法是explain plan for，其次就是set autotrace on；

3. 如果想观察到某条SQL有多条执行计划的情况，应使用dbms_xplan.display_cursor指定SQL_ID和CHILD_NUMBER的方法和查看指定SQL_ID语句的AWR报告的方法；

4. 如果SQL中含有多函数，函数中套有SQL等多层递归调用，想准确分析，应使用10046跟踪事件的方法；

       5. 要想获取表被访问的次数，应使用设置会话选项为statistics_level=all后，用dbms_xplan.display_cursor(null,null,’allstats last’)查看执行计划的方法；


上述方法中，哪些才能获得真实的执行计划？

顾名思义，只有被真正执行过的SQL，其所对应的执行计划才是真实的，反之，则不能保证你所看到的执行计划与真实执行时所采用的执行计划相同。
因此，上述方法中
1、explain for ... 
不能保证获取到真正的执行计划。

而剩下的四种方法：
2、set autotrace on
3、 dbms_xplan.display_cursor
4、 10046 trace跟踪
5、 awrsqrpt.sql
都可以获得真实的执行计划。但有一点需要注意，如果SQL中包含有绑定变量，那么用set autotrace on方法所获取到的执行计划，也有可能不是真实的。