runstats脚本分析不同的sql资源使用
runstats是《Oracle Database 9i/10g/11g编程艺术：深入数据库体系结构》作者编写的一个工具，能对做同一件事情的两个不同方法进行比较，得出孰优孰劣的结果。我们只需要提供两个不同的方法，余下的事情都由runstats负责。runstats只负责测量3个要素：
- 墙上时钟或耗时时间：知道墙上时钟或耗时时间很有用，不过这不是最重要的信息。
- 系统统计结果：会并排地i显示每个方法做某件事（如执行一个解析调用）的次数，并展示出两者之差
- 闩定（latching）：这是这个报告的关键输出。
要使用runstats，需要能访问几个V$视图，并创建一个表来存储统计结果，还要创建runstats包。为此，需要访问4个V$表（就是那些神奇的动态性能表）：V$STATNAME、V$MYSTAT和V$LATCH和V$TIMER。这四个表其实是别名，真正对象的名称应为V_$STATNAME、V_$MYSTAT、 V_$LATCH、 V_$TIMER，并且都是在sys账户下。如果其他账户要访问这四张表， 需要进行授权。
下面是安装方法:
1.设置权限
在你需要执行runstats包的用户赋予相关权限
grant SELECT on SYS.v_$statname to test ;
grant SELECT on SYS.v_$mystat to test ;
grant SELECT on SYS.v_$latch to test ;
grant SELECT on SYS.v_$timer to test ;
2.创建视图
--在test账户下创建视图
create or replace view stats
as select 'STAT...' || a.name name, b.value
from SYS.v_$statname a, SYS.v_$mystat b
where a.statistic# = b.statistic#
union all
select 'LATCH.' || name, gets
from SYS.v_$latch
union all
select 'STAT...Elapsed Time', hsecs from SYS.v_$timer;
3.创建收集信息的表
--创建信息收集表
create global temporary table run_stats
( runid varchar2(15),
name varchar2(80),
value int )
on commit preserve rows;
4.创建runstats包
create or replace package runstats_pkg
as
procedure rs_start;
procedure rs_middle;
procedure rs_stop( p_difference_threshold in number default 0 );
end;
/
4.创建runstats包体
--创建包体
create or replace package body runstats_pkg
as
g_start number;
g_run1 number;
g_run2 number;
procedure rs_start
is
begin
delete from run_stats;
insert into run_stats
select 'before', stats.* from stats;
g_start := dbms_utility.get_cpu_time;
end;
procedure rs_middle
is
begin
g_run1 := (dbms_utility.get_cpu_time-g_start);
insert into run_stats
select 'after 1', stats.* from stats;
g_start := dbms_utility.get_cpu_time;
end;
procedure rs_stop(p_difference_threshold in number default 0)
is
begin
g_run2 := (dbms_utility.get_cpu_time-g_start);
dbms_output.put_line
( 'Run1 ran in ' || g_run1 || ' cpu hsecs' );
dbms_output.put_line
( 'Run2 ran in ' || g_run2 || ' cpu hsecs' );
if ( g_run2 <> 0 )
then
dbms_output.put_line
( 'run 1 ran in ' || round(g_run1/g_run2*100,2) ||
'% of the time' );
end if;
dbms_output.put_line( chr(9) );
insert into run_stats
select 'after 2', stats.* from stats;
dbms_output.put_line
( rpad( 'Name', 30 ) || lpad( 'Run1', 12 ) ||
lpad( 'Run2', 12 ) || lpad( 'Diff', 12 ) );
for x in
( select rpad( a.name, 30 ) ||
to_char( b.value-a.value, '999,999,999' ) ||
to_char( c.value-b.value, '999,999,999' ) ||
to_char( ( (c.value-b.value)-(b.value-a.value)),
'999,999,999' ) data
from run_stats a, run_stats b, run_stats c
where a.name = b.name
and b.name = c.name
and a.runid = 'before'
and b.runid = 'after 1'
and c.runid = 'after 2'
and abs( (c.value-b.value) - (b.value-a.value) )
> p_difference_threshold
order by abs( (c.value-b.value)-(b.value-a.value))
) loop
dbms_output.put_line( x.data );
end loop;
dbms_output.put_line( chr(9) );
dbms_output.put_line
( 'Run1 latches total versus runs -- difference and pct' );
dbms_output.put_line
( lpad( 'Run1', 12 ) || lpad( 'Run2', 12 ) ||
lpad( 'Diff', 12 ) || lpad( 'Pct', 10 ) );
for x in
( select to_char( run1, '999,999,999' ) ||
to_char( run2, '999,999,999' ) ||
to_char( diff, '999,999,999' ) ||
to_char( round( run1/decode( run2, 0,
to_number(0), run2) *100,2 ), '99,999.99' ) || '%' data
from ( select sum(b.value-a.value) run1, sum(c.value-b.value) run2,
sum( (c.value-b.value)-(b.value-a.value)) diff
from run_stats a, run_stats b, run_stats c
where a.name = b.name
and b.name = c.name
and a.runid = 'before'
and b.runid = 'after 1'
and c.runid = 'after 2'
and a.name like 'LATCH%'
)
) loop
dbms_output.put_line( x.data );
end loop;
end;
end;
/
5.测试runstats脚本
runstats脚本分为三步
第一.执行runstats_pkg.rs_start开始调用,然后执行你第一段sql
第二.执行完成第一段sql后面执行runstats_pkg.rs_middle,然后执行你第二段的sql
第三.执行rs_stop,完成之后就会打印出你第一段sql和第二段sql的资源使用统计信息.
下面开始测试:
第一步:
SQL> set serveroutput on
SQL> exec runstats_pkg.rs_start
PL/SQL procedure successfully completed.
SQL> select count(*) from tobj;
COUNT(*)
----------
14921
第二步:
SQL> exec runstats_pkg.rs_middle
PL/SQL procedure successfully completed.
SQL> select /*+ full(tobj) */ count(*) from tobj;
COUNT(*)
----------
14921
第三步:
SQL> exec runstats_pkg.rs_stop
Run1 ran in 1 cpu hsecs
Run2 ran in 2 cpu hsecs
run 1 ran in 50% of the time
Name Run1 Run2 Diff
STAT...Requests to/from client 6 7 1
LATCH.threshold alerts latch 0 1 1
STAT...physical write total IO 0 1 1
STAT...db block gets direct 0 1 1
STAT...physical writes 0 1 1
STAT...physical writes direct 0 1 1
STAT...physical write IO reque 0 1 1
STAT...physical writes non che 0 1 1
STAT...physical writes direct 0 1 1
STAT...index scans kdiixs1 5 6 1
STAT...sql area purged 0 1 1
STAT...sql area evicted 0 1 1
STAT...cursor authentications 2 1 -1
STAT...parse count (failures) 0 1 1
LATCH.session switching 2 1 -1
LATCH.ksuosstats global area 3 2 -1
LATCH.object stats modificatio 0 1 1
LATCH.enqueues 1 2 1
LATCH.ksv allocation latch 1 0 -1
LATCH.queued dump request 0 1 1
LATCH.KMG MMAN ready and start 7 8 1
LATCH.cache buffers lru chain 1 2 1
LATCH.active checkpoint queue 7 8 1
LATCH.begin backup scn array 0 1 1
LATCH.dml lock allocation 1 0 -1
LATCH.KTF sga latch 0 1 1
LATCH.Change Notification Hash 7 8 1
LATCH.sort extent pool 1 0 -1
LATCH.deferred cleanup latch 1 0 -1
LATCH.shared pool sim alloc 2 1 -1
LATCH.session timer 7 8 1
LATCH.cp sga latch 1 0 -1
LATCH.ncodef allocation latch 1 0 -1
LATCH.ASM network state latch 1 0 -1
STAT...DB time 3 4 1
STAT...recursive cpu usage 3 1 -2
STAT...free buffer requested 10 12 2
STAT...calls to kcmgas 1 3 2
STAT...active txn count during 0 2 2
STAT...cleanout - number of kt 0 2 2
STAT...lob writes unaligned 0 2 2
STAT...session cursor cache co 2 0 -2
STAT...parse time elapsed 1 3 2
STAT...SQL*Net roundtrips to/f 5 7 2
LATCH.mostly latch-free SCN 20 22 2
LATCH.lgwr LWN SCN 20 22 2
LATCH.Consistent RBA 20 22 2
LATCH.object queue header heap 15 17 2
LATCH.LGWR NS Write 2 0 -2
LATCH.managed standby latch 2 0 -2
LATCH.alert log latch 2 0 -2
LATCH.shared pool simulator 8 6 -2
LATCH.PL/SQL warning settings 3 5 2
STAT...lob writes 0 3 3
STAT...user calls 7 10 3
STAT...index fetch by key 10 7 -3
STAT...HSC Heap Segment Block 16 19 3
LATCH.session allocation 4 1 -3
LATCH.session idle bit 18 21 3
LATCH.active service list 10 13 3
LATCH.simulator hash latch 22 25 3
STAT...enqueue requests 11 14 3
STAT...enqueue releases 11 14 3
LATCH.FAL Queue 3 0 -3
LATCH.archive control 3 0 -3
LATCH.loader state object free 0 4 4
LATCH.row cache objects 243 247 4
STAT...consistent changes 64 60 -4
STAT...table scans (short tabl 0 4 4
LATCH.tablespace key chain 1 6 5
STAT...non-idle wait count 5 10 5
LATCH.undo global data 8 13 5
STAT...calls to get snapshot s 23 29 6
LATCH.shared pool 263 269 6
LATCH.ASM db client latch 42 48 6
LATCH.space background task la 30 36 6
LATCH.redo allocation 61 67 6
LATCH.archive process latch 6 0 -6
STAT...session cursor cache hi 11 18 7
STAT...deferred (CURRENT) bloc 1 10 9
STAT...rows fetched via callba 9 0 -9
STAT...execute count 16 25 9
STAT...calls to kcmgcs 25 35 10
STAT...parse count (total) 16 26 10
LATCH.DML lock allocation 3 13 10
LATCH.OS process allocation 40 50 10
STAT...opened cursors cumulati 13 23 10
STAT...commit cleanouts 1 12 11
LATCH.redo writing 97 108 11
STAT...commit cleanouts succes 1 12 11
STAT...consistent gets - exami 34 22 -12
LATCH.object queue header oper 22 34 12
STAT...recursive calls 89 106 17
STAT...table fetch by rowid 31 10 -21
STAT...db block gets from cach 1 22 21
STAT...redo entries 18 43 25
STAT...db block changes 84 111 27
LATCH.enqueue hash chains 80 111 31
LATCH.channel operations paren 241 277 36
STAT...db block gets from cach 65 105 40
STAT...db block gets 65 106 41
STAT...buffer is not pinned co 73 31 -42
LATCH.messages 316 362 46
LATCH.checkpoint queue latch 729 833 104
STAT...consistent gets from ca 104 257 153
STAT...consistent gets 104 257 153
STAT...no work - consistent re 61 215 154
STAT...consistent gets from ca 70 235 165
STAT...session logical reads 169 363 194
STAT...table scan blocks gotte 0 199 199
LATCH.SQL memory manager worka 1,410 1,611 201
STAT...Elapsed Time 2,053 2,497 444
LATCH.cache buffers chains 453 971 518
STAT...bytes sent via SQL*Net 1,285 1,845 560
STAT...bytes received via SQL* 1,560 2,351 791
STAT...undo change vector size 4,148 5,708 1,560
STAT...physical write bytes 0 8,192 8,192
STAT...cell physical IO interc 0 8,192 8,192
STAT...physical write total by 0 8,192 8,192
STAT...redo size for direct wr 0 8,228 8,228
STAT...table scan rows gotten 0 14,929 14,929
STAT...redo size 5,768 21,416 15,648
STAT...session pga memory max 0 917,504 917,504
STAT...session pga memory 0 1,114,112 1,114,112
STAT...logical read bytes from 1,384,448 2,965,504 1,581,056
Run1 latches total versus runs -- difference and pct Run1 Run2 Diff Pct 4,418 5,436 1,018 81.27% PL/SQL procedure successfully completed.