wait event for switch log file command

[hrishy]

Hi All

The above wait event appears consistantly in my statspack report.

My log file sizes are 5Mb (samllish in size) i have two groups with two members each.They are not on raid 5.Any ideas how do we get rid of this wait event.



regards
Hrishy

[pando]

5MB... so small, create more groups or increase in size, i would do both

[hrishy]

Hi Pando

Thanks for the sugestion..I was thiking of somethin more reasonable like 20 -25 Mb..I will do this and see wheather i can elimnate this wait event.

regards
Hrishy

[DaPi]

I would certainly increase the number of groups.

As for the size, remember that you normally want some log switching: it allows the archlog to be generated (e.g. for standby) and forces a checkpoint, which would speed up an instance recovery. You may also want to limit the size to ease transmission of archlogs over the network. So the question here is: how frequently do you get a log switch with a 5MB file size? (and of course: how long is the wait? is it significant?)

[hrishy]

Hi Dapi

The archive logs switch at a half an hour interval.I saw this through the alert logs.The wait event si significant as revealed by the statspack report.


I dont have a standby for this instance.

Ideally i would have prefered a logswitch every hour.

I will try both i guess adding one more group and increasing the size of the members to 20Mb

any other suggestions thoughts are most welcome

regards
Hrishy

[DaPi]

Originally posted by hrishy
The wait event si significant as revealed by the statspack report.

Significant? (Remember, something has to be in the "top five waits". If your db & application are brilliantly tuned, with all data & parsed SQL in buffers, log switches & control file writes are about the only things left to wait on! ) So my question stands - how long do you wait for a switch?

[hrishy]

Hi Dapi

Yes i was checking the DB infact a routine performance review and this wait event showed up in the top 5 wait events along with controlm file parallel write.

The reason for control file parallel write being in the top 5 wait event was all the control files are placed on the same disk.

The next significant wait event was the one for log file switch thats when i started diggin in deeper.And started looking at the log file switch rates and the sizes of the logfiles.

what do u think ?

regards
Hrishy

[abhaysk]

Hrishy :

I think still u havent answered DAPI's

So my question stands - how long do you wait for a switch?

[hrishy]

Hi Abhay Dapi

I dont have exact number but in the top 5 wait events .

The wait event corresponding to log file switch and control file parallel write contribute 52% and 10% respectively.

regards
Hrishy

[marist89]

If you're switching every 30 minutes, I would question why you are waiting that much. I would carefully look at smaller timeslices in your statspack reports (maybe 5 minutes or so) to see what is really going on.