Latch Waits
Hi,
It appears to be that my LatchWaits are excessive as displayed by
the output of "tbstat -p". The database situation is this: We are
doing lots of selects during data-entry and by this I mean that we
are looking up as many as 30 tables per user for verification and
for insert/update purposes. This is happening for 30-50 users
picking at the system continuously. Additionally, the output phase
of the operation has approximately 10 users hitting 30-40 tables
and creating multiple temp tables in the process to speed up the
queries.
Now to the question, FINALLY.
If the latch waits are excessive (as they appear to me) what can I
do to decrease them? I am assumming, of course, that in reducing
the number of waits, the users will experience an increase in
throughput (although this MAY not be correct). What parameters
can I tune (read "Should I tune") to help this situation?
tbstat -p --->
dskreads pagreads bufreads %cached dskwrits pagwrits ufwrits %cached
1845092 1872221 88858140 97.92 58384 174583 12991941 99.55
isamtot open start read write rewrite delete commit rollbk
83452234 44066 5184946 34798223 5915604 1628049 61440 4516160 21
ovtbls ovlock ovuser ovbuff usercpu syscpu numckpts flushes
0 0 0 0 12977.03 2057.19 80 189
bufwaits lokwaits lockreqs deadlks dltouts lchwaits ckpwaits compress
258 0 48992292 0 0 30988 252 123486
TBCONFIG SETTING:
RESIDENT 0USERS 150
TRANSACTIONS 150
LOCKS 100000
BUFFERS 20000TBLSPACES 2000
CHUNKS 8
DBSPACES 8
PHYSBUFF 128
LOGBUFF 8LOGSMAX 10
CLEANERS 3
SHMBASE 0x0
CKPTINTVL 600
LRUS 24
LRU_MAX_DIRTY 20
LRU_MIN_DIRTY 10
LTXHWM 70
LTXEHWM 80
semmni 8192
semmns 8192
semmnu 360
shmmni 800
shmseg 120
Any help would more than greatly appreciated.
Keith Howard
Service Systems Enterprises, Inc.
(708) 297-2300x332