The 'redo log space requests' statistics counts messages sent to LGWR by
foreground processes (only). There are several types of messages, most of which
do not in fact signify a lack of space in the log buffer. The statistic name is
a hang over from version 6. From version 7.3 onwards a better way of checking
whether the log buffer might be too small has been to look for 'log buffer
space' waits in V$SYSTEM_EVENT. There is a lot more to be aware of about this,
but you can get all the other information on the Ixora web site. It is in
several places so I suggest you use the search box (you may even get a hit on
the text of this message which should be in the answers archive by then).

Incidentally, 'redo writes' is the number of write I/O requests issued by LGWR
to the operating system. It is the 'redo entries' statistic that counts the
number of times that foregrounds have allocated space in the log buffer.
Therefore, if you really do want to compute a ratio then 'redo entries' would be
a much more meaningful denominator than 'redo writes'.


-----Original Message-----
Sent: Saturday, 2 September 2000 0:01


I am using the freeware version of Toad to look at some performance ratios.
I am confused by the one they list as Redo Space Wait Raio. I captured the
query that computes this figure -

select (sw.value)*100/lw.value
from v$sysstat sw, v$sysstat lw
where sw.name='redo log space requests'
and lw.name='redo writes';

It does not seem to me that either of these values have anything to do with
waits, but are the number of occurances of the events. Could anyone help me
to understand what the 'redo log space requests' and 'redo writes' are or
why this ratio is so high on my Db (3.7)? Even better, could you tell me
where I could find the definitions of the columns in the v$sysstat table for
future use?