Re: Help me tuning this wait event:log file sync

Some new information: i check the statspack report and find there is about 20 commits/second in may, and now there is 40 commits/second these days. I do not know whether this is too much. THis is a busy oltp system,and the trend is that there will be more and more transactions per second, painful

And from statspack, the average redo size for one transactoin is 2K, Load Profile ~~~~~~~~~~~~ Per Second Per Transaction --------------- --------------- Redo size: 82,547.28 2,247.16 But from the result of sar: oracle_at_main-db1$sar -d 2 2

SunOS main-db1 5.7 Generic_106541-14 sun4u 07/13/02

22:07:33 device %busy avque r+w/s blks/s avwait avserv

22:08:35____ssd0,e____100____20.7___199____3189__0.0___104.4 This is the

      most busy time in the day for the system, and this device (ssd0,e
      ) is for the redo log volumn, 3189blocks/200=16blocks/second, that
      means 8KBytes per write.This is some small write, but still much
      greate than the 2k per transaction. why? other wasted log because
      of the too large log_buffer_size?

(there should be no read for redo log, right, unless it is being archived, so all write).

Thanks for your help:)

--
an oracle fan, 
an oracle beginner

an oracle fan, 
an oracle beginner

Posted via dBforums
http://dbforums.com