Oracle常用的性能诊断语句

--2.查找前10条性能差的SQL语句(磁盘读取较大，缺少索引或语句不合理)SELECT *  FROM (SELECT PARSING_USER_ID, EXECUTIONS,               SORTS,               COMMAND_TYPE,               DISK_READS,               sql_text          FROM v$sqlarea         ORDER BY disk_reads DESC) WHERE ROWNUM < 10;-- 查询消耗资源最多的SQLSELECT sql_text, hash_value, executions, buffer_gets, disk_reads, parse_calls FROM V$SQLAREAWHERE buffer_gets > 10000000 OR disk_reads > 100000ORDER BY buffer_gets + 100 * disk_reads DESC;--附SELECT EXECUTIONS, DISK_READS, BUFFER_GETS      , ROUND((BUFFER_GETS-DISK_READS) / BUFFER_GETS, 2) Hit_radio     , TRUNC(DISK_READS / EXECUTIONS) Reads_per_run     , SQL_TEXTFROM  V$SQLAREA WHERE EXECUTIONS > 0 AND DISK_READS > 0 and BUFFER_GETS > 0    AND (BUFFER_GETS-DISK_READS) / BUFFER_GETS < 0.8 ORDER BY 5 DESC;-- 查看某条SQL语句的资源消耗　SELECT HASH_VALUE, BUFFER_GETS, DISK_READS, EXECUTIONS, PARSE_CALLS  FROM  V$SQLAREA WHERE HASH_VALUE = 228801498   AND ADDRESS = HEXTORAW('CBD8E4B0');-- 运行时间很长的SQL（session_longops视图显示运行超过6秒的操作。包括备份，恢复，统计信息收集，查询等等）select b.USERNAME, b.SID, b.SERIAL#, b.START_TIME, b.LAST_UPDATE_TIME       , round(sofar * 100 / totalwork, 0) || '%' as progress       , a.sql_text from v$sqlarea a     join v$session_longops b on a.SQL_ID = b.SQL_ID--where b.TIME_REMAINING != 0        -- TIME_REMAINING：预计完成操作的剩余时间(秒) --  and b.USERNAME = 'sccot'-- 最近10分钟最消耗CPU的SQL语句： select sql_text from (        select sql_id,count(*) as cn        from v$active_session_history         where sample_time > sysdate - 10/24/60             and session_type <> 'BACKGROUND'            and SESSION_STATE = 'ON CPU'        group by sql_id        order by cn desc    ) ash, v$sql swhere ash.sql_id=s.sql_id;-- 最近10分钟最消耗IO的SQL语句：select sql_text from (        select sql_id,count(*) as cn        from v$active_session_history         where sample_time > sysdate - 10/24/60             and session_type <> 'BACKGROUND'            and WAIT_CLASS='User I/O'        group by sql_id        order by cn desc    ) ash, v$sql swhere ash.sql_id=s.sql_id;-- 补充1，根据时间找出响应时间较长的SQL：   select t.SQL_TEXT, t.SQL_FULLTEXT, t.ELAPSED_TIME, t.LAST_LOAD_TIME, t.*    from v$sql t    where t.LAST_LOAD_TIME > to_char(sysdate - 10/(24*60), 'yyyy-mm-dd/hh24:mi:ss') --and t.sql_text like '%LSHSXM%'     order by t.ELAPSED_TIME desc;-- 补充2，查找硬解析严重的SQL：select max(sql_id), substr(sql_text,0,100), count(1)from v$sql where executions < 10 group by substr(sql_text,0,100) having count(1) > 1000order by count(1) desc;--3.统计行数或空间占用较大的表select   t.table_name, t.num_rows, t.last_analyzed, t.temporary, t.*from     user_tables twhere t.num_rows > 0order by t.num_rows desc-- 普通表数据大小排序select segment_name,  trunc(bytes/1024/1024)from user_segments where segment_type = 'TABLE'order by bytes desc;-- 含LOB类型表的数据大小SELECT A.TABLE_NAME,        A.COLUMN_NAME,        trunc(B.BYTES / 1024 / 1024),        B.SEGMENT_NAME,        B.SEGMENT_TYPE,        B.TABLESPACE_NAME,        B.BLOCKS,        B.EXTENTS   FROM USER_LOBS A, USER_SEGMENTS B WHERE A.SEGMENT_NAME = B.SEGMENT_NAME ORDER BY B.BYTES DESC;-- 数据表总的数据大小占用select  rowSize + nvl(lobSize, 0) dataSize, t1.*, t2.*from (          select segment_name,  trunc(bytes/1024/1024) rowSize          from user_segments           where segment_type = 'TABLE'    ) t1 left join (             SELECT A.TABLE_NAME,                  A.COLUMN_NAME,                  trunc(B.BYTES / 1024 / 1024) lobSize,                  B.SEGMENT_NAME,                  B.SEGMENT_TYPE,                  B.TABLESPACE_NAME,                  B.BLOCKS,                  B.EXTENTS             FROM USER_LOBS A, USER_SEGMENTS B           WHERE A.SEGMENT_NAME = B.SEGMENT_NAME    ) t2 on t1.segment_name = t2.table_nameorder by rowSize + nvl(lobSize, 0) desc补充，查询该表实际使用的blocks：   SELECT COUNT(DISTINCT DBMS_ROWID.ROWID_BLOCK_NUMBER(rowid) || DBMS_ROWID.ROWID_RELATIVE_FNO(rowid)) AS Used_Blocks    FROM table_name;--4.查看Oracle内存参数配置  select *from   v$parameter twhere  t.NAME in (              'memory_max_target', 'memory_target', 'sga_max_size', 'sga_target', 'pga_aggregate_target', 'cpu_count'              , 'db_cache_size', 'shared_pool_size', 'large_pool_size', 'java_pool_size', 'streams_pool_size', 'log_buffer'              , 'db_2k_cache_size', 'db_4k_cache_size', 'db_8k_cache_size', 'db_16k_cache_size', 'db_32k_cache_size'       );-- 查询SGA区的分配细节及使用情况select t1.pool, t1.reserved_size_mb, t2.used_size_mb, t1.reserved_size_mb - t2.used_size_mb AS unused_size_mbfrom (       select t.POOL, trunc(sum(t.BYTES)/1024/1024) reserved_size_mb        from v$sgastat t        where t.POOL is not null        group by t.POOL    ) t1    left join (      select t.POOL, trunc(sum(t.BYTES)/1024/1024) used_size_mb        from v$sgastat t       where t.pool is not null and t.name != 'free memory'       group by t.POOL    ) t2 on t1.pool = t2.poolunion allselect t.name, trunc(sum(t.BYTES)/1024/1024) reserved_size_mb, null, null  from v$sgastat t where t.pool is null group by t.NAMEorder by  2 desc;-- 查看系统的缓存命中率与软解析率         SELECT 'Cache hit ratio' as name, (1-(physical_reads/(db_block_gets+consistent_gets)))*100 value FROM v$buffer_pool_statisticsunion all  select 'Soft parse ratio', 100-100*(a.value/b.value)  from v$sysstat a, v$sysstat b  Where  a.name='parse count (hard)' and b.name='parse count (total)'; --5.查询锁表的语句select 'kill -9 '||PS.SPID,     'alter system kill session '''||vs.sid||','||vs.serial#||''';',     /*       DECODE(V$LOCK.TYPE, 'TM', 'TABLE LOCK', 'TX', 'ROW LOCK', NULL) LOCK_LEVEL,*/       Decode(VL.LOCKED_MODE,0,'[0] none',                      1,'[1] null 空',                      2,'[2] Row-S 行共用(RS)：共用表鎖，sub share ',                      3,'[3] Row-X 行獨佔(RX)：用於行的修改，sub exclusive ',                      4,'[4] Share 共用鎖(S)：阻止其他DML操作，share',                      5,'[5] S/Row-X 共用行獨佔(SRX)：阻止其他事務操作，share/sub exclusive ',                      6,'[6] exclusive 獨佔(X)：獨立訪問使用，exclusive ',                      '['||VL.LOCKED_MODE||'] Other Lock') LockMode,    PS.SPID,OS_USER_NAME,VS.PROGRAM,VS.MACHINE,ORACLE_USERNAME,OBJECT_NAME,vs.LOGON_TIME ,   Vs.status,vs.MODULE, NVL(currentSql.SQL_TEXT, prevSql.SQL_TEXT) as sql_textfrom  V$LOCKED_OBJECT VL      join DBA_OBJECTS OB on VL.OBJECT_ID = OB.OBJECT_ID      join V$SESSION VS on VL.SESSION_ID = VS.SID       join v$process PS on PS.ADDR = VS.PADDR      --left join v$sql currentSql on vs.SQL_HASH_VALUE = currentSql.HASH_VALUE and vs.SQL_ADDRESS = currentSql.ADDRESS      --left join v$sql prevSql on vs.PREV_HASH_VALUE = prevSql.HASH_VALUE and vs.PREV_SQL_ADDR = prevSql.ADDRESS      left join v$sql currentSql on vs.SQL_ID = currentSql.SQL_ID and vs.SQL_Child_Number = currentSql.Child_Number       left join v$sql prevSql on vs.Prev_SQL_ID = prevSql.Sql_Id and vs.PREV_Child_Number = prevSql.Child_Numberorder by vs.logon_time;--6.产生kill会话的语句 select A.SID,       B.SPID,       A.SERIAL#,       a.lockwait,       A.USERNAME,       A.OSUSER,       a.logon_time,       a.last_call_et / 3600 LAST_HOUR,       A.STATUS,       'orakill ' || sid || ' ' || spid HOST_COMMAND,       'alter system kill session ''' || A.sid || ',' || A.SERIAL# || '''' SQL_COMMAND  from v$session A, V$PROCESS B where A.PADDR = B.ADDR   AND SID > 6;--7.查看IO情况select     df.name 文件名,    fs.phyrds 读次数,    fs.phywrts 写次数,    (fs.readtim/decode(fs.phyrds,0,-1,fs.phyrds)) 读时间,    (fs.writetim/decode(fs.phywrts,0,-1,fs.phywrts)) 写时间from  v$datafile df,    v$filestat fswhere df.file#=fs.file#order by df.name;--8.查看表空间情况SELECT UPPER(F.TABLESPACE_NAME) "表空间名",        D.TOT_GROOTTE_MB "表空间大小(M)",        D.TOT_GROOTTE_MB - F.TOTAL_BYTES "已使用空间(M)",        TO_CHAR(ROUND((D.TOT_GROOTTE_MB - F.TOTAL_BYTES) / D.TOT_GROOTTE_MB * 100, 2), '990.99') "使用比",        F.TOTAL_BYTES "空闲空间(M)",        F.MAX_BYTES "最大块(M)"  FROM (SELECT TABLESPACE_NAME,         ROUND(SUM(BYTES) / (1024 * 1024), 2) TOTAL_BYTES,         ROUND(MAX(BYTES) / (1024 * 1024), 2) MAX_BYTES         FROM SYS.DBA_FREE_SPACE GROUP BY TABLESPACE_NAME) F,      (SELECT DD.TABLESPACE_NAME,ROUND(SUM(DD.BYTES) / (1024 * 1024), 2) TOT_GROOTTE_MB         FROM SYS.DBA_DATA_FILES DD         GROUP BY DD.TABLESPACE_NAME) D WHERE D.TABLESPACE_NAME = F.TABLESPACE_NAME ORDER BY 4 DESC;--9.根据unix上Top命令看到的PID,查找对应的SQlSELECT P.pid pid,       S.sid sid,       P.spid spid,       S.username username,       S.osuser osname,       P.serial# S_#,       P.terminal,       P.program program,       P.background,       S.status,       RTRIM(SUBSTR(a.sql_text, 1, 80)) SQL  FROM v$process P, v$session S, v$sqlarea A WHERE P.addr = s.paddr   AND S.sql_address = a.address(+)   AND P.spid LIKE '%CPU最高的进程对应的PID%';-- 数据库及实例的状态信息select name, log_mode, open_mode, flashback_on, supplemental_log_data_min, platform_name from v$database;select instance_number, instance_name, host_name, version, startup_time, status from v$instance;--清空缓存(共享池和数据缓存)，Command window执行：alter system flush shared_pool;alter system flush buffer_cache;--索引信息查看： select t.table_owner, t.index_name, t.blevel, t.leaf_blocks       , t.last_analyzed, t.distinct_keys, t.num_rows, t.sample_sizefrom user_indexes t where t.table_name = upper('workitem');--analyze table 更新统计信息：analyze table my_table compute statistics; analyze index idx_name compute statistics;--OREXEC DBMS_STATS.gather_table_stats(ownname => 'socct', tabname =>'WORKITEM', estimate_percent =>100, cascade =>true, method_opt => 'for all columns size auto');/*参数说明：一、 estimate_percent ：抽样统计百分比二、 cascade ：为 false 时 , 只对表进行统计分析；为 true 时 , 同时对表和索引进行统计分析 .三、 method_opt ：　 method_opt: 决定 histograms 信息是怎样被统计的 .method_opt 的取值如下 :　　 for all columns: 统计所有列的 histograms.　　 for all indexed columns: 统计所有 indexed 列的 histograms.　　 for all hidden columns: 统计你看不到列的 histograms　　 for columns <list> SIZE <N> | REPEAT | AUTO | SKEWONLY　　　　  SIZE ：统计指定列的 histograms.N 的取值范围 [1,254];  　　　　REPEAT 上次统计过的 histograms;　　　　  AUTO 由 oracle 决定 N 的大小 ;  四、 degree ：并发性，可调值为， 8/16/24/32...*/   --删除数据后，释放数据空间alter table my_table enable row movement;alter table my_table shrink space cascade;alter table my_table disable row movement;oralter table table_name move; -- 执行此操作后，需要重建索引select t.NUM_ROWS, t.TEMPORARY, t.last_analyzed, t.* from user_tables t where t.TABLE_NAME = upper('gspauresult_tkk0107');select t.num_rows, t.distinct_keys, t.sample_size, t.last_analyzed, t.* from user_indexes t where t.table_name = upper('tkk007');select t.* from user_ind_columns t where t.TABLE_NAME = upper('gspauresult_tkk0107')order by t.INDEX_NAME, t.COLUMN_POSITION;select dbms_metadata.get_ddl('TABLE',upper('gspauresult_tkk0107')) from dual;select dbms_metadata.get_ddl('INDEX',upper('idx_gspauresult_tkk0107')) from dual;select dbms_metadata.get_ddl('CONSTRAINT',upper('tkk0107')) from dual;select dbms_metadata.get_ddl('REF_CONSTRAINT',upper('tkk0107')) from dual;select * from user_tables t order by dbms_random.random;获取指定SQL在内存中的执行计划：select sql_id,child_number,sql_text from v$sql where sql_text like 'select count(1) from emp a where a.dept_no =%';select * from table(DBMS_XPLAN.DISPLAY_CURSOR('sql_id',0)); 1、PL/SQL Developer 中使用F52、explain plan for select count(1) from emp a where a.dept_no=5;     select * from table(dbms_xplan.display());3、sqlplus中使用 set autotrace traceonly exp;