Hi Elardus,
SMF Type 90 (Subtype 11) simply gives the OPT names: SMF90OPO = old OPT
parmlib member name, SMF90OPN = new OPT parmlib member name. These names
don't tell you much about the values actually specified in the old and new
OPT members. :-(
The lack of OPT information in SMF is indeed unfortunate. As Martin Packer
implied, automated performance analysis tools cannot easily check for
common OPT problems if the data is not easily available!
In a legacy version of CPExpert, the "SRM Component" checked
IPS/ICS/OPT/TSOKEY parameters by reading the relevant members in
sys1.parmlib. This approach was prone to user error, since the actual
sys1.parmlib members had to be available to the system on which CPExpert
was running, but the system being evaluated was not necessarily the system
on which CPExpert ran! Even then, operator changes to OPT specifications
would not always be reflected in sys1.parmlib members.
With modern z/OS environments, most of IBM subsystems are
"self-documenting" so automated performance analysis tools can easily build
a "model" of the environment being analyzed.
With WLM, most of the relevant information is available in TYPE 72, and SMF
is "self documenting" with regard to service class period information,
resource group information, etc. There still is the missing workload
classification information, but most information necessary for analyzing
WLM performance problems is available in SMF. Specifications related to
PR/SM, IRD, Blocked Workload, HiperDispatch Management Mode, and other
system parameter specifications are available in SMF. As a result,
automated performance analysis products can happily read SMF data and have
an understanding of detailed performance issues in these z/OS performance
areas.
Similarly, a legacy version of the CPExpert CICS Component had to read the
System Initialization Table (SIT) parameters directly from the SIT (and any
operator overrides had to be explicitly provided). This was annoying to
users, particularly for those users with a large number of unique (not
cloned) CICS regions. Since CICS MVS/ESA Release 4.1, most (and all
important) SIT parameters are provided in SMF Type 110 records, and the
CICS Development Team continues to add SIT values to the Type 110 records
as new features are added to CICS. Automated performance analysis products
can happily read SMF Type 110 data and have an understanding of detailed
performance issues in CICS performance areas.
All important DB2 parameters and specifications are available in SMF
(particularly if Trace Type 106 is available).
A complete model of the DASD I/O environment now can be constructed
entirely from SMF data.
WebSphere MQ parameter specifications are available in SMF (and detailed
queue information is available if SMF Type 116 is available).
The bottom line of this is that IBM has been moving extensively toward
providing "self documenting" data in SMF.
The glaring lack of complete performance-related OPT information is the one
major area in which "self documenting" specifications have not been made
available in SMF.
It almost seems to be a hit-or-miss situation as to whether OPT
specifications are available in SMF!. As examples, there is complete
information in SMF Type 70 records about OPT specifications for some new
features (such as OPT specifications controlling Blocked Workload or
HiperDispatch Management Mode), but there is no OPT information in SMF
about other new features (such as ZAAPAWMT or ZIIPAWMT). It is
disheartening to know that performance problems could easily be identified
and solutions recommended automatically, if only the required OPT
information were available in SMF!
Placing the remaining performance-related OPT information in the SMF Type
70 records would significantly improve automated performance analysis and
would have almost zero effect on SMF file size.
Regards,
Don
******
Don Deese, Computer Management Sciences, Inc.
Voice: (703) 922-7027 Fax: (703) 922-7305
http://www.cpexpert.com
******
"Elardus Engelbrecht" <[EMAIL PROTECTED]> wrote in message
news: <[EMAIL PROTECTED]>...
[ This is a resend atttempt, apparently my first post never went to IBM-MAIN ]
Martin Packer wrote:
>All the OPT parms and WLM policy details and any dynamic overrides of
those ought to be in an SMF record. "Self documenting" would be a GOOD thing.
Indeed. Look for SMF type 90, subtype 11. Look for SMF90TID.
Groete / Greetings
Elardus Engelbrecht
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
