Uneven distribution of latches

From: Boris Dali <boris_dali_at_yahoo.ca>
Date: Fri, 24 Dec 2004 06:30:14 -0800 (PST)
Message-ID: <20041224143014.14071.qmail@web41411.mail.yahoo.com>

The subject says it all.
The distribution below (from v$latch_children for latch# 157 - library cache):

CHILD# GETS MISSES SLEEPS WAIT_TIME ------ ---------- ---------- ---------- ----------

     2   16597807      72387        333    1674350
     1   34475788     203781       1254    4801328
     7   27065517     241493       1284    5753591
     8   33897775     440298       1718   14570006
     4   60799550     676421       1833   11250799
    10   62099645     922629       2881   11083064
     6   52733583     818465       3384   21322498
     3   56602708     883077       9150   49110401
     9   65188664    1275753      10214   64507786
    11  117460816    3227510      24170  133460244
     5 3643740682 3001597195   23020516 6.2108E+12

So child latch 5 is the one getting hit the most by far. I know exactly the reason/query that gets soft-reparsed over and over, but is there **anything** that can be done from a system/DB configuration perfspective to:
- ease the load on CPUs pegging at a 100%

make the distribution more even (would it improve anything, e.g. decrease response time and/or overall system load?)? Why child latch 5?

Thanks,
Boris Dali.

Do you Yahoo!?
Send holiday email and support a worthy cause. Do good. http://celebrity.mail.yahoo.com

--
http://www.freelists.org/webpage/oracle-l

Received on Fri Dec 24 2004 - 08:25:40 CST

Uneven distribution of latches - what's next?