Redo Log times ?

Question

During peak times my redo log times go sky high.. like 104.65ms .. How can I tune this to be lower ?

Answer 1

The "sync" is the time it takes for a process to send a message to the log writer, the log writer to notice the message and write any pending log buffer to disc, and then send a message back to the initiating process.
When the return message is sent, the process may be at the bottom of a long run queue, and take some time to get to the top of the queue and notice the message. This is most likely to happen when your system gets busy - especially with a large number of concurrent processes that are enthusiastically consuming the CPU.
An increase in the "sync" time may be about CPU and process counts and have nothing to do with the ability of the log writer to do its job.
Unfortunately, the log writer write times aren't captured properly in 9i, so it's hard to tell if the log writer really is starting to collapse under pressure; but if you've got sync times of 1/10 second a typical job isn't going to see it as much of a threat unless it's a pseudo-batch job that is inserting and committing 1,000,000 rows one at a time. So you have to ask if there are any jobs which are visibly much slower under load, rather than depending on the totalled log file sync time as an indicator - a check of v$session_event may give you a view on whether any session is suffering particularly badly.
You might want to read this from Kevin Closson:
http://kevinclosson.wordpress.com/2007/07/21/manly-men-only-use-solid-state-disk-for-redo-logging-lgwr-io-is-simple-but-not-lgwr-processing/
Regards
Jonathan Lewis
http://jonathanlewis.wordpress.com
http://www.jlcomp.demon.co.uk