The latch wait list and its latches are explained on page 27 of my book. The
others are all associated with parallel execution. These miss rates are probably
not of concern. It is of the nature of parallel execution that several processes
tend to want the same latches at the same time. In any event, as with all
latching, it is the number of sleeps that you should be concerned with, not the
number of misses.


-----Original Message-----
Sent: Thursday, 31 August 2000 23:20


I guess that I am applying the standard broad brush that latches with a
%miss rate of over 1% are bad and something should be done to fix them.  I
certainly could be in error.

In Oracle Performance Tuning - Tips and Techniques by Niemiec, on page 182
he states that "If the number of immediate misses is more than 1 percent
.... you may have a contention problem."  Therefore, I have assumed that I
have a contention problem.

Do you have an idea of what these latches mean?  Are the miss rates OK and
nothing to worry about?  We have seen some intermittent bad response times
in production so I am trying to tune it up as best I can.


-----Original Message-----
Sent: Thursday, August 31, 2000 8:51 AM


What makes you think that there is a problem here?


-----Original Message-----
Sent: Thursday, 31 August 2000 23:09


AIX 4.3.2
Oracle 8.1.6
No OPS

Does anyone know what the following latches mean and more importantly how to
fix them?

Latch                       Gets Misses %misses
--------------------       ----- ------ -------
latch wait list              5754   368  6.39%
parallel query alloc buffer 27083 10653 39.33%
parallel query stats         1003   453 45.16%
process queue                9979   113  1.13%
process queue reference   4157244 61439  1.47%
queryserver freelists       16348   522  3.19%