Log Buffer Tuning

[parisdba]

hi,
My problem is as follows,

1)the redo buffer allocation retries and log space request in the V$systat increases constantly by about 2 (each of them with max of upto 12 ) durring each check point. is there any scope for tuning here and if so how.

2)I am new to this databse and it has been configured with log buffer of 10 MB and redo log file size of 30 mb(as per advice of oracle support and i/o speed is not much of concern since we have a dedicated filer for it) . I was searchiing the net for optimal sizes of log buffer and found that optimal sizes to be around 1-2 Mb could anyone explain to me the basis of sizing the log buffer and redo log files

 

[vivekm]

On the point of size of the redo log buffer, the theory behind the sizing conflicts with your findings. When the log writer flushes out the records to the redo log file, normally, memory read and writes are faster than disk read/writes. This leads to two conclusions:
1. Having a larger redo log buffer will result in less read/write conflict.
2. The speed of I/O does not depend on the size of the redo log files. Speed of disk writes will only be affected adversely due to fragmentation.

I would allocate space to the redo buffer according to Oracle's recommendation initially : "default setting is four times the maximum data block size for the host operating system." and then review at a later stage for contention.

Regards,
Vivek