What fields from IBM SMF72.3 is MICS using to arrive at the fields in WLM

Products

MICS Resource Management

Issue/Introduction

We're not clear on how WLMSEC file is calculated if transaction goes to a different address space like DB2 which is derived from SMF Type 72 subtype 3 record.

We want to be able to use file WLM_SEC to report the information.

The two fields of interest are:

Actual Response time 25.4 milliseconds and TRX Rate 2295 (Transactions Per Second ended)

Response Time shown in RMF III when issuing the "RTD report-class" command see below

What fields from IBM SMF72.3 does MICS use
to arrive at the fields in WLM_SEC ?

Environment

Release:
Component: MICS

Resolution

You asked what fields from IBM SMF72.3 MICS is using to arrive at the fields below in WLM_SEC .

SECTCPTM - Total CPU Time
SECPSRCP - Transactions Per Second
SECAVNCP - Avg Transaction Execution Time
SECAVXET - Avg Transaction Execution Time
SECAVTET - Avg Transaction Elapsed Time

I'm going to start with the last 4 because they are easiest to explain.

In general, you can determine exactly how all data elements are derived by examining the DIC.TEXT members, DERIVATION section.

2) SECPSRCP - Transactions per second
DERIVATION from DIC.TEXT(SECPSRCP): SECPSRCP =SECTORCP / DURATION

---where DERIVATION section for SECTORCP says: contains contents of field R723CRCP (number of completed transactions)

DURATION is the duration of MICS observation in seconds (RMF Interval (DETAIL), or longer in DAYS, WEEKS, etc.

3) SECAVNCP - Avg Transactions Execution Time
DERIVATION from DIC.TEXT(SECAVNCP): SECAVNCP = SECTOXET / SECTONCP

---where DERIVATION section for SECTOXET says: contains contents of field R723CXET (total transaction execution time )

SECTONCP says: contains contents of field R723CNCP (number of transactions completed execution phase)

4) SECAVXET - Avg Transactions Execution Time
DERIVATION from DIC.TEXT(SECAVXET): SECAVNCP = SECTOXET / SECTORCP

---where DERIVATION section for SECTOXET says: contains contents of field R723CXET (total transaction execution time)

SECTORCP says: contains contents of field R723CRCP (number of transactions completed)

5) SECAVTET - Avg Transactions Elapsed Time
DERIVATION from DIC.TEXT(SECAVTET): SECAVTET = SECTOTET / SECTORCP

---where DERIVATION section for SECTOTET says: contains contents of field R723CTET (total transaction elapsed time)

SECTORCP says: contains contents of field R723CRCP (number of transactions completed)

1) SECTCPTM: Total CPU Time - this one is quite complicated. Number 1, IBM does not record CPU times in the 72-3 record. Instead it stores service units, and we must convert the service units to CPU time based on other field values.

This element "Total CPU Time" is basically the sum of: CP Core TCB time + CP Core SRB time + normalized zAAP core CPU time + normalized zIIP core CPU time.

1) SECTCPTM: Total CPU Time - the DERIVATION says:

ELEMENT DERIVATION: SECTCPTM is the sum of SECCPUTM, SECSRBTM, SECZAPTM, and SECSUPTM.

So, you must look at the DERIVATION for these 4 elements:

a) SECCPUTM: TCB CPU Time on CP

SECTOCPU
SECCPUTM = ----------------------------------- - (SECZAPTM + SECSUPTM)
WLMSCCPU*WLMSUCPU

SECTOCPU - TCB CPU Service Units on CP - field R723CCPU
WLMSCCPU - Service Coefficient - CPU - field R723MCPU
WLMSUCPU - Effec. SU:CPU-second Conversion Factor - 16,000,000 / field R723MADJ

SECZAPTM - CPU Time on zAAP (Normalized), computed as shown below:

R723CIFA * (R723NFFI/256)
------------------------------
R723MCPU * (16000000/R723MADJ)

SECSUPTM - CPU Time on zIIP (Normalized), computed as shown below:

R723CSUP * (R723NFFS/256)
------------------------------
R723MCPU * (16000000/R723MADJ)

b) SECSRBTM: SRB CPU Time

SECSRBTM - SRB CPUT Time computed as shown below:

SECTOSRB / (WLMSCSRB*WLMSUCPU)

SECTOSRB - TCB CPU Service Units on CP - field R723CSRB (total SRB service units)
WLMSCSRB - Service Coefficient - SRB - field R723MSRB
WLMSUCPU - Effec. SU:CPU-second Conversion Factor - 16,000,000 / field R723MADJ

c) and d) zAAP and zIIP CPU times are computed as shown above. .