Re: Database performance at COMMON

Rob

Have you tried running a monitor against this stuff - in Ops Nav - visual explain can give you lots of help. It would tell you the access method, and that might be enough.

Exists is always, at least almost always, the most efficient thing - like a SETLL *EQ

Vern

-------------- Original message --------------
From: rob@xxxxxxxxx

I am going to COMMON. My main mission this time is database performance.
Which, if any, of the sessions would best have helped me with this email I
just received from a coworker?


There have been some UPDATE SQL statements taking a long time in the GDI
database, probably because its so large. Particularly, against files like
ECL. The WHERE ... IN appears to be the culprit. A "WHERE ... IN" type
statement doesn't seem to take advantage of reusable data very well, even
if the table being compared is fairly small. I suspect its like the
difference between scanning all data in a table vs. using a CHAIN. See
below for a similar but much more efficient statement.

Took 2 hours on HQ:
update gdidivf/ecl set
lnet=(select pfct1 from darren/hmi264254 where
ecl.lord=hmi264254.ord and ecl.lline=hmi264254.line)
where (lord,lline) in
(select lord,lline from darren/hmi264254)

Took 20 seconds:
update gdidivf/ecl set
lnet=(select pfct1 from darren/hmi264254 where
ecl.lord=hmi264254.ord and ecl.lline=hmi264254.line)
where exists
(select pfct1 from darren/hmi264254 where
ecl.lord=hmi264254.ord and ecl.lline=hmi264254.line)


P.S.
Please note that the selection criteria following the 'where exists' is
identical to the value being set in the actual update statement. What you
are selecting, pfct1 in this example, is not important, but it is easier
just to copy paste this way.


By the way "HQ" is short for GDIHQ, which is an lpar with ninety 70GB disk
drives running at 60% full

% CPU used . . . . . . . : 34.9 Auxiliary storage:
% DB capability . . . . : 20.2 System ASP . . . . . . : 5433 G
Elapsed time . . . . . . : 00:00:14 % system ASP used . . :
60.0069
Jobs in system . . . . . : 31474 Total . . . . . . . . : 5433 G
% perm addresses . . . . : .347 Current unprotect used : 96810
M
% temp addresses . . . . : 7.740 Maximum unprotect . . : 501608
M

System Pool Reserved Max -----DB----- ---Non-DB---
Pool Size (M) Size (M) Active Fault Pages Fault Pages
1 4901.22 641.37 +++++ .0 .0 6.1 6.1
2 21317.14 212.66 1497 .4 271.5 398.3 1493
3 396.32 .08 39 .0 8.5 7.6 42.6
4 5944.83 .00 735 3.6 13.8 13.5 23.9
5 7072.73 .00 165 .9 4.2 3.0 7.6

DSPHDWRSC *PRC
Resource Type-model Status Text
CEC01 9406-MMA Operational Main Card Enclosure
PN03 28D4 Operational System Control Panel
MP13 53CF Operational System Processor Card
MP14 53CF Operational System Processor Card
MP15 53CF Operational System Processor Card
MP16 53CF Operational System Processor Card
MP17 53CF Operational System Processor Card
MP18 53CF Operational System Processor Card
MP19 53CF Operational System Processor Card
MP20 53CF Operational System Processor Card
PV03 52AD Operational Processor Capacity Card
SP04 294E Operational Service Processor Card
SP05 294E Operational Service Processor Card

Rob Berendt
--
Group Dekko Services, LLC
Dept 01.073
Dock 108
6928N 400E
Kendallville, IN 46755
http://www.dekko.com

--
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing list
To post a message email: MIDRANGE-L@xxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: http://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxx
Before posting, please take a moment to review the archives
at http://archive.midrange.com/midrange-l.