Action Trace Example

This example shows how to use subtype 1 trace data with subtype 2 service class period data to understand what actions SRM is taking, why those actions are taken, and which work is affected.

The table below shows the subtype 1 trace data output:

CLASS    P     SPI   LPI  ACTION
NRBATCH  2     .02   .03  3610  rv_hsk_inc_mpl
TSO      1     .10   .11  2630  tdh_rem_prt
TSO      2    1.84  1.84   270  pa_rec_cand
TSO      1     .10   .11   880  pa_pro_rdon_cand
TSO      2    1.84  1.84   620  pa_pmuo_rec
APPC     1    1.10  1.10   960  pa_pro_unc_sec_don
APPC     3    1.10  1.10   960  pa_pro_unc_sec_don
NRBATCH  1    2.23  3.23   960  pa_pro_unc_sec_don
TSO      1     .10   .11   940  pa_pro_unc_don
TSO      2    1.08  1.08   750  pa_pro_incp_rec

The data has the following headings:

CLASS

The service class name.

P

The service period number within the service class.

SPI

The sysplex performance index for the service class period traced.

LPI

The local performance index for the service class period traced.

ACTION

The action code and mnemonic.

The table below shows the subtype 2 service class period output.

CLASS    P I N  SC GT    SPI   LPI DP  MPLI MPLO  SWPT
APPC     1 2 N -99 SRT  1.10  1.10 251    0  999   488
APPC     2 2 N -99 SRT     0     0 251    0  999     0
APPC     3 2 N -72 SRT  1.10  1.10 251    1  999 76206
APPC     4 3 N -99 VEL     0     0 249    1  999     0
NRBATCH  1 3 N -99 LRT  2.16  3.04 249    1   11     0
NRBATCH  2 4 Y -98 LRT   .02   .03 247    2  999     0
NRBATCH  3 6 Y -99 DIS   .81   .81 192    0   14     0
OFFBATCH 1 6 Y -99 DIS   .81   .81 192    0   14     0
TSO      1 2 Y -64 SRT   .10   .11 251    1   11     0
TSO      2 2 Y   0 SRT  1.84  1.84 251    1  999  1464
TSO      3 2 Y  -2 SRT   .91   .93 247    2   12  2440
TSO      4 3 Y -99 VEL   .26   .26 245    7  999     0
$SRMDISC 1 6 Y -99 DIS   .81   .81 192    0   14     0 

The subtype 2 service class policy period output has the following headings:

CLASS

The service class name.

P

The service period number within the service class.

I

The service class period's importance, with 1 being highest.

N

The “needs help” indicator. Y means the service class period needs help and N means the service class period doesn't need help.

SC

The skip clock.

GT

One of the following goal types:

• SRT - short response time goals
• LRT - long response time goals
• VEL - velocity goals
• DIS - discretionary goals.

SPI

The sysplex performance index for the service class period traced.

LPI

The local performance index for the service class period traced.

DP

Dispatching priority

MPLI

MPL-in target

MPLO

MPL-out target

SWPT

Swap protect time

Interpreting the Trace Data

Based on the information in these subtypes, SRM took the following actions:

• rv_hsk_inc_mpl, reverse housekeeping, to increase the MPL targets for service class NRBATCH, period 2.

  This action code indicates the service class period was using significantly more MPL slots than guaranteed by its MPL control so its MPL in target was increased.

• tdh_rem_prt, time driven housekeeping, removing the swap protect time target from first period TSO.

  First period TSO is easily meeting its goals and was assessed to not need swap protect time.

• pa_rec_cand indicates TSO period 2 was chosen as the receiver.

  This shows the start of a series of actions where SRM is trying to improve the performance of TSO period 2 by increasing its dispatching priority. SRM selected TSO period 2 as the receiver because it had the worst performance index, 1.84, of all the work at the highest importance defined, 2. Service class NRBATCH, period 1, had a worse performance index, 2.16, but it also had a lower importance, 3, so it would be chosen as a receiver after TSO period 2 which had a higher importance and was not meeting goals.

• pa_pro_rdon_cand indicates that TSO period 1 has been selected as the processor donor to be assessed.

  SRM selected TSO period 1 as the first donor candidate because it had the best performance index of all service class periods that were running at a dispatching priority higher than or equal to the receiver's dispatching priority.

• pa_pmuo_rec indicates that the first dispatching priority move to be assessed is to move TSO period 2 up to a higher dispatching priority.
• The next several pa_pro_unc_sec_don trace entries show the other service class periods whose processor access will be affected by the move up of TSO period 2 even though their priority remains unchanged. These service class periods were all secondary donors. TSO period 1 was the primary donor. This is indicated by the pa_pro_unc_don action.
• pa_pro_incp_rec shows that the primary receiver, TSO period 2, received a dispatching priority increase.

The sysplex and local performance index projections are shown with the final action for each service class period affected.