dg库日志应用慢参数调整

dg库日志应用慢参数调整

 

1> 日志应用是否使用了并行;-并行度看主机的cpu个数

1 alter database recover managed standby database parallel 8 disconnect from session;

 

2> 调整并行恢复参数 parallel_execution_message_size 最大值65536；增加size需要更加多的share pool。

调整参数时，需要确定shared pool的使用情况，避免4031错误

1 show parameter parallel_execution_message_size
2 show parameter disk_asynch_io
3 show parameter parallel_execution_message_size
4 
5 alter system set parallel_execution_message_size=32768 scope=spfile;
6 alter system set disk_asynch_io=true scope=spfile;

 

3> 是否使用异步io；通常异步io能提供更好的io性能；aix hp 貌似默认是开启异步io的

1 alter system set filesystemio_options=setall;
2 show parameter filesystemio_options

 

4> db_block_checking --块内数据逻辑检查

在dg库上取消该参数，能提供更好的性能，经过测试能提升约2倍速度  

 1 alter system set db_block_checking=false scope=spfile;
 2 
 3  
 4 startup nomount;
 5 alter database mount standby database;
 6 
 7 --------------------------------------------
 8 
 9 alter system set parallel_execution_message_size=32768 scope=spfile; 
10 alter system set filesystemio_options=setall scope=spfile;  
11 alter system set disk_asynch_io=true scope=spfile;
12 alter database recover managed standby database cancel;
13 
14 shutdown immediate;
15 startup 
16 
17 alter database recover managed standby database parallel 4 disconnect from session using current logfile; 

       

极端情况下DG加快恢复速度===在standby端执行，千万不可在primary端调整

 1 alter system set parallel_execution_message_size=32768 scope=spfile;   16384
 2 alter system set filesystemio_options=setall scope=spfile; 默认 none
 3 alter system set disk_asynch_io=true scope=spfile;                   默认就是true
 4 
 5 alter system set db_lost_write_protect=typical scope=spfile  ;       默认是full
 6 
 7 alter system set db_block_checksum=false scope=spfile ;  默认是TYPICAL
 8 
 9 alter system set DB_BLOCK_CHECKING=false scope=spfile ;         默认是false
10 
11 alter system set db_writer_processes=8 scope=spfile;                  默认是6个
12 
13 shutdown immediate;
14 
15 startup nomount;
16 alter database mount standby database;
17 alter database recover managed standby database parallel 4 disconnect from session; ?----并行度根据CPU核数*2设定

 

另外shared_pool size过小也会影响应用速度，请遇到DG延迟的时候 可适当根据安装规范检查，防止shared pool过小。（如最小要保证2G以上）

 

dg库日志应用性能监控

 1 set lines 200 pages 2000
 2 col process format a8
 3 col spid format a8
 4 col event format a50 tru
 5 col SIW format 999999
 6 
 7 select to_char(sysdate,'DD-MON-YYYY HH24:MI:SS') "Current time"
 8  ,s.process
 9  , p.spid
10  , substr(s.program, -6) PROC
11  , s.event
12  , s.p1
13  , s.p2
14  , s.p3
15  , s.seconds_in_wait SIW
16  , s.seq#
17 from v$session s, v$process p
18 where p.addr = s.paddr and (s.program like '%MRP%' or s.program like '%PR0%' or s.program like '%DBW%' or s.program like '%CKPT%')
19 order by s.process
20 
21 /