INTRODUCTION TO QUERY/400 Chapter 7

Work Management
& Performance
Chapter 14
Overview

The AS/400
combines the pieces
of Managing Devices,
Working with Jobs,
and Working with
Subsystems to run
programs

Pieces fit together –
performance testing
What makes a job run?






Devices
6 necessary things:
Job Descriptions
Job Classes
Job Queues
Subsystems with:
 Routing entry(s)
 Job queue entries – batch SBS
 Workstation entry(s) – interactive SBS
 Communication entry(s) – communication SBS
A valid user ID or password
How the AS/400 runs

Interactive Jobs





WS device known
Valid sign-on
SBS with job
description and class
WSE
Routing Entry

Batch Jobs

How batch jobs are
started:




SBMJOB
Communication PGM
start request
Auto-start jobs
Pre-start jobs
How the AS/400 runs

Communication Jobs



A hybrid between interactive and batch
jobs
Remote system start
Communication entry needed
Subsystems

QBASE and QCTL

SYSOPR should:




QCMN
QINTER
QBATCH
QSPL
Performance


AS/400 has ability to constantly “tune”
resources while jobs are running
Document all efforts
Automatic Performance Tuning

Two system values

QDYNPTYSCD


QPFRADJ


prevents gobbling resources
manages pool memory allocations
Manual monitoring is still a good idea
Why Manually Monitor?

DPS is quick to reset changes priorities after
they have been lowered



SYSOPR may want it low
SYSOPR may want a PGM to be a “hog”
Sometimes user response-time complaints
are due to transmission time across a busy
or large LAN. (AS/400 performance tools
don’t address this because it takes place
outside the AS/400)
Emergencies – What do I do?

An operator should be concerned with 2
areas:

Overall CPU %


Over 100% indicated by ++++
CPU % for individual jobs
WRKACTJOB
Work with Active Jobs
CPU %:
.0
Elapsed time:
Type options, press Enter.
2=Change
3=Hold
4=End
8=Work with spooled files
Opt
_
_
_
_
_
_
_
_
_
Subsystem/Job
QBASE
QBATCH
QCMN
QCTL
QSYSSCD
QINTER
QPADEV0004
QPGMR
QSERVER
User
QSYS
QSYS
QSYS
QSYS
QPGMR
QSYS
DAWSON#M
QSYS
QSYS
00:00:00
07/03/99
Active jobs:
112
5=Work with
6=Release
13=Disconnect ...
Type
SBS
SBS
SBS
SBS
BCH
SBS
INT
SBS
SBS
CPU %
.0
.0
.0
.0
.0
.0
.0
.0
.0
BIGBLUE
17:47:11
7=Display message
Function
PGM-QEZSCNEP
CMD-WRKACTJOB
Status
DEQW
DEQW
DEQW
DEQW
EVTW
DEQW
RUN
DEQW
DEQW
More...
Parameters or command
===>_________________________________________________________________
F3=Exit
F5=Refresh
F7=Find
F10=Restart statistics
F11=Display elapsed data
F12=Cancel
F23=More options
F24=More keys
Figure
12-14:
Jobs and subsystems active on an AS/400. From the WRKACTJOB
Why is CPU % high?

3 Reasons



Programs are complex
Page faulting is causing a lot of extra disk
IO
Overall load on the system (# of jobs) is
too high
Recommendations Vs. Reality

IBM recommends:



below 70% for single-processors
Below 81% for a four-way processor
Systems without noticeable performance
problems will typically run in the 90% range

Remember: CPU utilization over 90% gets
hammered from excessive seize/lock conflicts
What Next?


CPU % over 90% - even if normal continue searching for problem within
the AS/400
CPU % good (approx 75%), hang-up
may be due to lost device on LAN
Response time problem + low CPU %
Suspect LAN
Work With Active Jobs
Example




Largest CPU % - F16 re-sequences by
specified column
Some programs need a lot of
“horsepower”
Confirm big jobs are necessary (while
holding - 3)
If necessary, lower the job priority
(higher #)
Problem still exists?

WRKSYSACT (Work System Activity)


Screen refreshes every five seconds
System tasks and their CPU % is displayed


Don’t normally appear on WRKACTJOB screen
Cancel the user job and the system tasks
will go with it
Emergency Aftermath



New priorities require time to take affect
Recovery period could cause worse damage
LAN failure – disconnect users, sessions
dumping logs and cleaning up after
themselves


lots of activity
Announcements – “Sign back on”

Performance will die – 15 to 45 minutes
TIPS


Develop guidelines for performance
emergencies
Always use Performance tuning



Monitor performance during busiest time
Modify system parameters (changing activity
levels or memory) during less busy times
REMEMBER: Monitoring is an ongoing
process

If the System Operator has the AS/400 properly
tuned, performance emergencies are less likely to
occur
TIPS (Cont.)

Performance Tools/400 has powerful datacollecting and reporting tools






STRPFRMON
ADDPFRCOL
ANZPFRDTA
PRTTNSRPT
Best/1 is a built-in capacity-planning product
Use SETOBJACC to preload entire files into
dedicated pools for certain kinds of
processing
AS/400 Servers



Tuned to favor batch processing, leave
them that way
A few WS will make the interactive CPU
% go to 10% and then nothing will run
well
IBM recommends less than 2% of the
interactive CPU %