监控数据库性能的SQL

geniiking

希望整理一个监控oracle性能的常用sql集锦，希望大家不断更加：

1. 监控事例的等待

    select event,sum(decode(wait_Time,0,0,1)) "rev",
   sum(decode(wait_Time,0,1,0)) "Curr",count(*) "Tot"
    from v$session_Wait
    group by event order by 4;

2. 回滚段的争用情况

    select name, waits, gets, waits/gets "Ratio"
    from v$rollstat a, v$rollname b
    where a.usn = b.usn;

3. 监控表空间的 I/O 比例

    select df.tablespace_name name,df.file_name "file",f.phyrds pyr,
    f.phyblkrd pbr,f.phywrts pyw, f.phyblkwrt pbw
    from v$filestat f, dba_data_files df
    where f.file# = df.file_id
    order by df.tablespace_name;

4. 监控文件系统的 I/O 比例

    select substr(a.file#,1,2) "#", substr(a.name,1,30) "Name",
    a.status, a.bytes, b.phyrds, b.phywrts
    from v$datafile a, v$filestat b
    where a.file# = b.file#;

5.在某个用户下找所有的索引

    select user_indexes.table_name, user_indexes.index_name,uniqueness, column_name
    from user_ind_columns, user_indexes
    where user_ind_columns.index_name = user_indexes.index_name
    and user_ind_columns.table_name = user_indexes.table_name
    order by user_indexes.table_type, user_indexes.table_name,
    user_indexes.index_name, column_position;

6. 监控 SGA 的命中率

    select a.value + b.value "logical_reads", c.value "phys_reads",
    round(100 * ((a.value+b.value)-c.value) / (a.value+b.value)) "BUFFER HIT RATIO"
    from v$sysstat a, v$sysstat b, v$sysstat c
    where a.statistic# = 38 and b.statistic# = 39
    and c.statistic# = 40;

7. 监控 SGA 中字典缓冲区的命中率

    select parameter, gets,Getmisses , getmisses/(gets+getmisses)*100 "miss ratio",
    (1-(sum(getmisses)/ (sum(gets)+sum(getmisses))))*100 "Hit ratio"
    from v$rowcache
    where gets+getmisses <>;0
    group by parameter, gets, getmisses;

8. 监控 SGA 中共享缓存区的命中率，应该小于1%

    select sum(pins) "Total Pins", sum(reloads) "Total Reloads",
    sum(reloads)/sum(pins) *100 libcache
    from v$librarycache;

    select sum(pinhits-reloads)/sum(pins) "hit radio",sum(reloads)/sum(pins) "reload percent"
    from v$librarycache;

9. 显示所有数据库对象的类别和大小

    select count(name) num_instances ,type ,sum(source_size) source_size ,
    sum(parsed_size) parsed_size ,sum(code_size) code_size ,sum(error_size) error_size,
    sum(source_size) +sum(parsed_size) +sum(code_size) +sum(error_size) size_required
    from dba_object_size
    group by type order by 2;

10. 监控 SGA 中重做日志缓存区的命中率，应该小于1%

    SELECT name, gets, misses, immediate_gets, immediate_misses,
    Decode(gets,0,0,misses/gets*100) ratio1,
    Decode(immediate_gets+immediate_misses,0,0,
    immediate_misses/(immediate_gets+immediate_misses)*100) ratio2
    FROM v$latch WHERE name IN ('redo allocation', 'redo copy');

11.  监控内存和硬盘的排序比率，最好使它小于 .10，增加 sort_area_size

    SELECT name, value FROM v$sysstat WHERE name IN ('sorts (memory)', 'sorts (disk)');


12. 监控当前数据库谁在运行什么SQL语句

    SELECT osuser, username, sql_text from v$session a, v$sqltext b
    where a.sql_address =b.address order by address, piece;

13. 监控字典缓冲区

    SELECT (SUM(PINS - RELOADS)) / SUM(PINS) "LIB CACHE" FROM V$LIBRARYCACHE;
    SELECT (SUM(GETS - GETMISSES - USAGE - FIXED)) / SUM(GETS) "ROW CACHE" FROM V$ROWCACHE;
    SELECT SUM(PINS) "EXECUTIONS", SUM(RELOADS) "CACHE MISSES WHILE EXECUTING" FROM V$LIBRARYCACHE;

    后者除以前者,此比率小于1%,接近0%为好。

    SELECT SUM(GETS) "DICTIONARY GETS",SUM(GETMISSES) "DICTIONARY CACHE GET MISSES"
    FROM V$ROWCACHE

14. 找ORACLE字符集

    select * from sys.props$ where name='NLS_CHARACTERSET';

15. 监控 MTS

    select busy/(busy+idle) "shared servers busy" from v$dispatcher;

    此值大于0.5时，参数需加大

    select sum(wait)/sum(totalq) "dispatcher waits" from v$queue where type='dispatcher';
    select count(*) from v$dispatcher;
    select servers_highwater from v$mts;

    servers_highwater接近mts_max_servers时，参数需加大

16. 碎片程度

    select tablespace_name,count(tablespace_name) from dba_free_space group by tablespace_name
    having count(tablespace_name)>;10;

    alter tablespace name coalesce;
    alter table name deallocate unused;

    create or replace view ts_blocks_v as
    select tablespace_name,block_id,bytes,blocks,'free space' segment_name from dba_free_space
    union all
    select tablespace_name,block_id,bytes,blocks,segment_name from dba_extents;

    select * from ts_blocks_v;

    select tablespace_name,sum(bytes),max(bytes),count(block_id) from dba_free_space
    group by tablespace_name;

    查看碎片程度高的表

    SELECT segment_name table_name , COUNT(*) extents
    FROM dba_segments WHERE owner NOT IN ('SYS', 'SYSTEM') GROUP BY segment_name
    HAVING COUNT(*) = (SELECT MAX( COUNT(*) ) FROM dba_segments GROUP BY segment_name);

17. 表、索引的存储情况检查

    select segment_name,sum(bytes),count(*) ext_quan from dba_extents where
    tablespace_name='&tablespace_name' and segment_type='TABLE' group by tablespace_name,segment_name;

    select segment_name,count(*) from dba_extents where segment_type='INDEX' and owner='&owner'
    group by segment_name;

18、找使用CPU多的用户session

    12是cpu used by this session

    select a.sid,spid,status,substr(a.program,1,40) prog,a.terminal,osuser,value/60/100 value
    from v$session a,v$process b,v$sesstat c
    where c.statistic#=12 and c.sid=a.sid and a.paddr=b.addr order by value desc;

ruknow

up

伯爵的Druid

请教楼主，在监控event waits中的 sql*net 算是什么等待？(我在toad中的dba 中的database monitor中观察到的)
是oracle的net服务等待？如何降低这样的waits？
----------------------
环境：
appserver 运行weblogic 8
dbserver 运行oracle 92
sun fire v880 solairs 8 0202

pearaep

有些老，但还不错。支持一下。

zhangyh123

原帖由 "伯爵的Druid" 发表：
请教楼主，在监控event waits中的 sql*net 算是什么等待？(我在toad中的dba 中的database monitor中观察到的)
是oracle的net服务等待？如何降低这样的waits？
----------------------
环境：
appserver 运行webl..........

这类 非空闲等待事件不需要你处理，没任何问题，

还有类似的比如 client message
pmon timer
rdbms ipc message 等
都没关系

伯爵的Druid

toad中dba dababase monitor的event waited中有以下指标
control file io
single block read
multi block read
direct path read
sql *net
file io
log write
楼上的意思,这些都没关系么?

natalie_ho

好帖子！

ycdch

不知你是从那里“整理”过来的，其中关于碎片那一段sql不对啊！！

PK强

不错哦支持楼主

zgahj

收藏，收藏