Take the 2-minute tour ×
Database Administrators Stack Exchange is a question and answer site for database professionals who wish to improve their database skills and learn from others in the community. It's 100% free, no registration required.

The wait event “log buffer space” according to the documentation can indicate that the log buffer is too small or the I/O is slow. We have a system seeing this event in which the redo buffer allocation retries is going up in V$SYSSTAT as observed with...

SELECT VALUE FROM V$SYSSTAT WHERE NAME = 'redo buffer allocation retries';  

We recently increased the LOG_BUFFER value from 8 MB to 16 MB with only a small reduction in the retries. It is still increasing at about 1,000 a day.

Should the LOG_BUFFER value be increased further?

Since the system is 64 bit I can take it up to 256 MB, but I don’t want to use the memory for this buffer if it won’t be beneficial and I don’t know if there are any downsides to increasing this value that significantly.

share|improve this question
    
on what storage configuration are the redologs written? One reason for the increase can be a small buffer, the other is slow logwriter.... –  ik_zelf Mar 9 '11 at 6:25
    
They are written to an HP EVA 4400 SAN with 24, 146 GB 15K drives in the applicable disk group. The storage is shared with other databases that do not experience this issue, but they also don't write as much redo. –  Leigh Riffel Mar 9 '11 at 15:03
    
Leigh,normally you want to have the redos on raid-10, preferably on dedicated disks, not on raid-5 and not shared with other used files. What raid level is in use here? –  ik_zelf Mar 9 '11 at 19:33
    
That is absolutely what I want, but not what I have. The number of drives is actually 40 and they are raid-5. –  Leigh Riffel Mar 10 '11 at 22:27
    
Related: asktom.oracle.com/pls/apex/… –  Leigh Riffel Nov 26 '13 at 21:03

1 Answer 1

up vote 3 down vote accepted

Take it all the way up. FWIW I have systems where 500m and even 1g ORLs are normal. Don't be afraid of this on modern hardware - set archive_lag_target to give you a reasonable log switch (I go for 230 seconds to keep it off 5-minute boundaries).

share|improve this answer
    
FYI, we recently dropped this to 75 seconds with no performance impact –  Gaius Apr 27 '11 at 21:25

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.