很典型的sql优化

项目中一个应用反映如下查询速度缓慢，查询整体下来50多秒，在昨天给处理完毕后记录下来给大家分享下：
select * from V_CIRCUIT$OC where (CIRCUITCODE like '%XXYY%') and belongcity in('XZ0116','XZ011699','XZ011654','XZ011640','XZ011641','XZ011642','XZ011643','XZ011644','XZ011645','XZ011646','XZ011648','XZ011649','XZ011650','XZ011651','XZ011652','XZ011653') order by circuitcode ,belongcity;
根据这个查询来看sql上来看,sql本身需要改进的地方不多，所以着重观察所选定的目标对象视图。这个视图的创建sql语句如下：
create VIEW "V_CIRCUIT$OC" ("CIRCUITCODE", "USERNAME", "STATION1", "STATION2", "STATIONA", "STATIONZ", "CITY1", "CITY2", "SERIAL", "X_PURPOSE", "CREATETIME", "USETIME", "RATE", "PROPERTY", "HIRECIRCUITCODE", "PATH", "EQUIPCODE_A", "NENAME_A", "FRAMESERIAL_A", "SLOTSERIAL_A", "PACKSERIAL_A", "PORTSERIALNO1", "SLOT1", "EQUIPCODE_Z", "NENAME_Z", "FRAMESERIAL_Z", "SLOTSERIAL_Z", "PACKSERIAL_Z", "PORTSERIALNO2", "SLOT2", "AREA", "SYSTEMCODE", "LEASER", "HIREDATE", "SCHEDULERID", "REQUISITIONID", "STATE", "REMARK", "UPDATEPERSON", "UPDATEDATE", "CIRCUITSERIAL", "CIRCUITLEVEL", "ZB_SCHEDULEID", "VENDOR", "USERENDTOEND", "DLDH", "ISMAIN", "ISLOCAL", "BELONGCITY", "EQUIPLABELA", "PORTNAMEA", "EQUIPLABELB", "PORTNAMEZ", "ZBDH", "SEQUENCEID", "DDFNAME_A", "DDFCODE_A", "DDFNAME_Z", "DDFCODE_Z", "PROVINCENAME", "JH_DDF_A", "JH_DDF_Z", "JH_MODULE_A", "JH_MODULE_Z", "JH_STATION_A", "JH_STATION_Z", "BELONG", "NEWPURPOSE", "NEWCIRCUIT", "CS_DDF_A", "CS_DDF_Z" ) AS
SELECT a."CIRCUITCODE",
a."USERNAME",
a."STATION1",
a."STATION2",
sa."STATIONCODE" stationa,
sz."STATIONCODE" stationz,
a."CITY1",
a."CITY2",
a."SERIAL",
a."X_PURPOSE",
a."CREATETIME",
a."USETIME",
a."RATE",
a."PROPERTY",
a."HIRECIRCUITCODE",
a."PATH",
ea."EQUIPCODE",
ea."NENAME",
qa."FRAMESERIAL",
qa."SLOTSERIAL",
qa."PACKSERIAL",
a."PORTSERIALNO1",
a."SLOT1",
ez."EQUIPCODE",
ez."NENAME",
qz."FRAMESERIAL",
qz."SLOTSERIAL",
qz."PACKSERIAL",
a."PORTSERIALNO2",
a."SLOT2",
a."AREA",
a."SYSTEMCODE",
a."LEASER",
a."HIREDATE",
a."SCHEDULERID",
a."REQUISITIONID",
a."STATE",
a."REMARK",
a."UPDATEPERSON",
a."UPDATEDATE",
a."CIRCUITSERIAL",
a."CIRCUITLEVEL",
a."ZB_SCHEDULEID",
a."VENDOR",
a."USERENDTOEND",
a."DLDH",
a."ISMAIN",
a."ISLOCAL",
a."BELONGCITY",
getEQUIPLabel(a.portserialno1) EQUIPLABELA,
getportlabel (a.portserialno1) port1,
getEQUIPLabel(a.portserialno2) EQUIPLABELB,
getportlabel (a.portserialno2) port2,
'' zbdh,
''sequenceid,
(SELECT ddfa.ddfportname FROM v_sdh_ddf_link ddfa WHERE ddfa.sdhptp = a.portserialno1 AND ROWNUM < 2) AS ddfname_a,
(SELECT ddfa.ddfptp
FROM v_sdh_ddf_link ddfa
WHERE ddfa.sdhptp = a.portserialno1 AND ROWNUM < 2) AS ddfcode_a,
(SELECT ddfa.ddfportname FROM v_sdh_ddf_link ddfa WHERE ddfa.sdhptp = a.portserialno2 AND ROWNUM < 2) AS ddfname_z,
(SELECT ddfa.ddfptp FROM v_sdh_ddf_link ddfa WHERE ddfa.sdhptp = a.portserialno2 AND ROWNUM < 2) AS ddfcode_z,
xt.xtxx provincename,
a.jh_ddf_a,
a.jh_ddf_z,
a.jh_module_a,
a.jh_module_z,
a.jh_station_a ,
a.jh_station_z,
a.belong,
a.newpurpose,
a.newcircuit,
a.cs_ddf_a,
a.cs_ddf_z
FROM circuit a LEFT JOIN (equiplogicport qa JOIN equipment ea ON (qa.equipcode =
ea.equipcode
)
JOIN station sa ON (ea.stationcode = sa.stationcode))
ON (qa.logicport = a.portserialno1)
LEFT JOIN (equiplogicport qz JOIN equipment ez ON (qz.equipcode =
ez.equipcode
)
JOIN station sz ON (ez.stationcode = sz.stationcode))
ON (qz.logicport = a.portserialno1)
JOIN xtbm xt ON (a.belongcity = xt.xtbm AND xt.xtbm<>'0' )

了解到v_sdh_ddf_link 也是一个创建视图，而且是动态的，所以查询慢的问题，先就从这个特殊视图的创建语句中开始。
在建立这个视图中的sql语句中，提取如下语句信息：
SELECT portserialno1 sdhptp, porttype1 t1, portserialno2 ddfptp, porttype2 t2,decode(porttype2,'ZY23010499',getOdfportlabel(portserialno2),getddfportlabel(portserialno2)) ddfportname
FROM matrix_stos a
WHERE a.porttype1 like 'ZY0307040_' AND (a.porttype2 = 'ZY23010499' OR a.porttype2 = 'ZY13010499')
UNION
SELECT portserialno2 sdhptp, porttype2 t1, portserialno1 ddfptp, porttype1 t2,decode(porttype1,'ZY23010499',getOdfportlabel(portserialno1 ),getddfportlabel(portserialno1 ))ddfportname
FROM matrix_stos a
WHERE a.porttype2 like 'Z’;
发现本身这个视图建立的查询很慢。看来性能故障就在此。
由于本库sqlplus的sql_trace无无法正常显示，所以采用了建立会话级别跟踪查询计划方式：
alter session set sql_trace=true;
运行该查询，在udump下生成hbcsdb_ora_17772.trc文件，然后采用tkprof进行格式化
tkprof hbcsdb_ora_17772.trc session.log
打开log文件
TKPROF: Release 10.2.0.1.0 - Production on M-PM-GM-FM-ZM-RM-; 4M-TM-B 12 15:30:30 2010
Copyright (c) 1982, 2005, Oracle. All rights reserved.
Trace file: hbcsdb_ora_17772.trc
Sort options: default
********************************************************************************
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
********************************************************************************
alter session set sql_trace=true

call count cpu elapsed disk query current rows
------- ------ -------- ---------- ---------- ---------- ---------- ----------
Parse 0 0.00 0.00 0 0 0 0
Execute 1 0.00 0.00 0 0 0 0
Fetch 0 0.00 0.00 0 0 0 0
------- ------ -------- ---------- ---------- ---------- ---------- ----------
total 1 0.00 0.00 0 0 0 0
Misses in library cache during parse: 0
Optimizer mode: ALL_ROWS
Parsing user id: 114
********************************************************************************
SELECT portserialno1 sdhptp, porttype1 t1, portserialno2 ddfptp, porttype2 t2,decode(porttype2
,'ZY23010499',getOdfportlabel(portserialno2),getddfportlabel(portserialno2)) ddfportname
...skipping...
TKPROF: Release 10.2.0.1.0 - Production on M-PM-GM-FM-ZM-RM-; 4M-TM-B 12 15:30:30 2010
Copyright (c) 1982, 2005, Oracle. All rights reserved.
Trace file: hbcsdb_ora_17772.trc
Sort options: default
********************************************************************************
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
********************************************************************************
alter session set sql_trace=true

call count cpu elapsed disk query current rows
------- ------ -------- ---------- ---------- ---------- ---------- ----------
Parse 0 0.00 0.00 0 0 0 0
Execute 1 0.00 0.00 0 0 0 0
Fetch 0 0.00 0.00 0 0 0 0
------- ------ -------- ---------- ---------- ---------- ---------- ----------
total 1 0.00 0.00 0 0 0 0
Misses in library cache during parse: 0
Optimizer mode: ALL_ROWS
Parsing user id: 114
********************************************************************************
SELECT portserialno1 sdhptp, porttype1 t1, portserialno2 ddfptp, porttype2 t2,decode(porttype2,'ZY23010499',getOdfportlabel(portserialno2),get
ddfportlabel(portserialno2)) ddfportname
FROM matrix_stos a
WHERE a.porttype1 like 'ZY0307040_' AND (a.porttype2 = 'ZY23010499' OR a.porttype2 = 'ZY13010499')
UNION
SELECT portserialno2 sdhptp, porttype2 t1, portserialno1 ddfptp, porttype1 t2,decode(porttype1,'ZY23010499',getOdfportlabel(portserialno1 ),ge
tddfportlabel(portserialno1 ))ddfportname
FROM matrix_stos a
WHERE a.porttype2 like 'ZY0307040_' AND (a.porttype1 = 'ZY23010499' OR a.porttype1 = 'ZY13010499')
call count cpu elapsed disk query current rows
------- ------ -------- ---------- ---------- ---------- ---------- ----------
Parse 1 0.01 0.00 0 0 0 0
Execute 1 0.00 0.00 0 0 0 0
Fetch 3839 3.24 3.40 0 1643 0 57561
------- ------ -------- ---------- ---------- ---------- ---------- ----------
total 3841 3.25 3.41 0 1643 0 57561
Misses in library cache during parse: 1
Optimizer mode: ALL_ROWS
Parsing user id: 114
Rows Row Source Operation
------- ---------------------------------------------------
57561 SORT UNIQUE (cr=404760 pr=0 pw=0 time=5361001 us)
57561 UNION-ALL (cr=404760 pr=0 pw=0 time=5182488 us)
53106 TABLE ACCESS FULL MATRIX_STOS (cr=822 pr=0 pw=0 time=53196 us)
4455 TABLE ACCESS FULL MATRIX_STOS (cr=821 pr=0 pw=0 time=5549 us)
********************************************************************************
SELECT '(' || D.EQUIPNAME || ')' || P.DDFDDMSERIAL || '-' || P.PORTSERIAL
FROM
EN_DDF D,EN_DDFPORT P WHERE P.EQUIPCODE = D.EQUIPCODE AND P.DDFPORTCODE =
:B1

call count cpu elapsed disk query current rows
------- ------ -------- ---------- ---------- ---------- ---------- ----------
Parse 1 0.00 0.00 0 0 0 0
Execute 57306 0.71 0.63 0 0 0 0
Fetch 57306 1.64 1.41 0 401360 0 57306
------- ------ -------- ---------- ---------- ---------- ---------- ----------
total 114613 2.35 2.04 0 401360 0 57306
query的值引起了我的注意。
Misses in library cache during parse: 1
Misses in library cache during execute: 1
Optimizer mode: ALL_ROWS
Parsing user id: 114 (recursive depth: 1)
********************************************************************************
SELECT '(' || D.EQUIPNAME || ')' || P.ODFODMSERIAL || '-' || P.PORTSERIAL
FROM
EN_ODF D,EN_ODFPORT P WHERE P.EQUIPCODE = D.EQUIPCODE AND P.ODFPORTCODE =
:B1

call count cpu elapsed disk query current rows
------- ------ -------- ---------- ---------- ---------- ---------- ----------
Parse 1 0.00 0.00 0 0 0 0
Execute 255 0.00 0.00 0 0 0 0
Fetch 255 0.00 0.00 0 1757 0 248
------- ------ -------- ---------- ---------- ---------- ---------- ----------
total 511 0.00 0.01 0 1757 0 248
Misses in library cache during parse: 1
Misses in library cache during execute: 1
Optimizer mode: ALL_ROWS
Parsing user id: 114 (recursive depth: 1)
********************************************************************************

上面的SELECT '(' || D.EQUIPNAME || ')' || P.DDFDDMSERIAL || '-' || P.PORTSERIAL
FROM
EN_DDF D,EN_DDFPORT P WHERE P.EQUIPCODE = D.EQUIPCODE AND P.DDFPORTCODE =
:B1 引起了我的注意，询问现场说是getDDFPortLabel 函数的值，该函数的定义如下：
FUNCTION getDDFPortLabel(portCode in VARCHAR2)
return VARCHAR2 AS
--thePort equiplogicport.portserial%type;
ret VARCHAR2(100);
BEGIN
select '(' || d.equipname || ')' || p.ddfddmserial || '-' || p.portserial into ret
from en_ddf d,en_ddfport p
where p.equipcode = d.equipcode
and p.ddfportcode = portCode;
return ret;
END;
对涉及到的表查看数据量信息：
SQL> select count(*) from EN_DDF;
COUNT(*)
----------
18556
SQL> select count(*) from en_ddfport;
COUNT(*)
----------
664347
通过:
selct index_name,column_name from user_ind_columns where table_name='EN_DDF';
selct index_name,column_name from user_ind_columns where table_name=upper('en_ddfport');
检查发现关键字索引都不存在。
所以就创建了equipcode、dfportcode、porttype1 、porttype2关键字作为索引字段；
根据oracle表连接性能查询来看，驱动表应该选取数据量小的表，而本程序中恰好犯了一个错误。
然后建议将两个表的连接顺序进行对换，即为：
select '(' || d.equipname || ')' || p.ddfddmserial || '-' || p.portserial into ret
from en_ddfport p,en_ddf d
where p.equipcode = d.equipcode
and p.ddfportcode = portCode;
注意：如果是CBO优化模式就没有这个规定
然后检查相关连接字段索引，对不存在的建立好。重新测试上面的查询，发现速度的确得到根本上的提升，哈哈，结果杂眼就出来，可行通过工具无法在这给显示截图。
总之，一句话：数据库的性能提升根本上还是在与程序的改善。