database writer kicking in

I was trying to get the answer from metalink also but i think this forum is much better that that

we are experincing database writer to be kicked in and stayed there 4 to 5 minutes some time. It is because of the log switch as when logswitch is applied then it applies the checkpoint and also the force the dbwr to clear the dirty buffer but my question is that why it stays for 4 to 5 minutes becasue
Oracle writes the dirty buffers to disk only on certain conditions:

- A shadow process must scan more than one-quarter of the db_block_buffer
parameter.
- Every three seconds.
- When a checkpoint is produced

but it seems that it is clearing the dirty buffer only at the point of check point not after every 3 seconds or because of first reason.

do you know that how can i find that it is clearing dirty buffer after 3 seconds.

what else i need to investigate for this problem to improve dbwr activity.

Thanks
UKDBA

You can also set the number of dirty blocks the database writer writes per batch by setting the
db_writer_dirty_batch parameter in the init.ora file.Also check the alert log file for any message "checkpoint incomplete"whic means the dbwr has not completed the checkpoint process.In such case add more log groups


cheese
anandkl

In alert log there isn't any error about checkpoint not completed.
well i was trying to get some info about db_writer_dirty_batch from metalink but couldn't find anything. Can you give me more details about it.

UKDBA