Support for the Workload License Charges (WLC) Pricing Model

This document describes recent RMF enhancements related to the Workload License Charges (WLC) pricing model.

Introduction
This section is a very short introduction into the Workload License Charges (WLC) pricing model. For more detailed information about WLC, please refer to www.ibm.com/servers/eserver/zseries/wlc_lm/

If you are interested in RMF specific information only, start with the section RMF Postprocessor Partition Data Report.

Workload License Charges
Prior to z/OS, IBM software products running on OS/390 were typically priced based on the computing capacity of the Central Processor Complex (CPC) on which the software was running.

With z/OS running on a zSeries 900, the new concept of Workload Licensed Charges (WLC) is used to manage the system. There are two categories of products under WLC.

Variable Workload License Charge (VWLC) products
The costs are based on the capacity defined for the product (less than the total CPC capacity).

In the figure above, you can see z/OS WebSphere is charged with 100 MSUs although the total CPC has a capacity of 320 MSUs.

Flat Workload License Charge (FWLC) products
The costs for these products are a flat rate, independent of the size of the CPC on which they run.

LPAR Resource Management
z/OS manages workloads to LPAR defined capacities, based on a four-hour rolling average. A workload can use more than an LPAR's defined capacity, as long as the average workload during the four-hour period does not exceed the LPAR's defined capacity.

RMF Postprocessor Partition Data Report
If you require data that is related to software pricing for long-term performance measurement, use the Partition Data Report section in the RMF CPU Activity Report. This report has been extended by SW pricing related data MSU DEF, MSU ACT and WLM CAPPING % (percentage of time when WLM capped the partition). It is invoked by REPORTS(CPU):

                                              p A R T i T i O N  D A T A  R E p O R T
                                                                                                                           PAGE    2
            z/OS V1R2                SYSTEM ID SYSF             START 12/07/2001-12.05.00  INTERVAL 000.59.59
                                     RPT VERSION V1R2 RMF       END   12/07/2001-13.05.00  CYCLE 1.000 SECONDS

MVS PARTITION NAME                    SYSF
IMAGE CAPACITY                          50
NUMBER OF CONFIGURED PARTITIONS         15
NUMBER OF PHYSICAL PROCESSORS            8
                   CP                    7
                   ICF                   1


WAIT COMPLETION                        MIX
DISPATCH INTERVAL                  DYNAMIC

--------- PARTITION DATA -----------------  -- LOGICAL PARTITION PROCESSOR DATA --   -- AVERAGE PROCESSOR UTILIZATION PERCENTAGES --
                   ----MSU----  -CAPPING--  PROCESSOR-  ----DISPATCH TIME DATA----   LOGICAL PROCESSORS  --- PHYSICAL PROCESSORS ---
NAME       S   WGT  DEF    ACT  DEF   WLM%  NUM   TYPE   EFFECTIVE       TOTAL       EFFECTIVE    TOTAL  LPAR MGMT  EFFECTIVE  TOTAL
VMA        A   200    0      7  NO     0.0    7   CP    00.12.09.017  00.13.30.675        2.89     3.22      0.32       2.89   3.22
SYSA       A   100    0      6  NO     0.0    5   CP    00.09.48.481  00.10.27.122        3.27     3.48      0.15       2.34   2.49
SYSB       A    50    0      5  NO     0.0    5   CP    00.08.21.036  00.09.04.989        2.78     3.03      0.17       1.99   2.16
SYSC       A    50    0      4  NO     0.0    5   CP    00.06.17.683  00.06.53.692        2.10     2.30      0.14       1.50   1.64
SYSD       A    32    0     29  NO     0.0  3.8   CP    00.53.54.897  00.54.13.253       23.34    23.47      0.07      12.84  12.91
SYSE       A    57   45     57  NO     0.0  4.0   CP    01.46.45.812  01.47.22.740       44.49    44.74      0.15      25.42  25.57
SYSF       A    60   50     74  NO     0.0  5.0   CP    02.20.16.068  02.20.59.620       46.76    47.00      0.17      33.40  33.57
KEY3       A    50    0      2  NO     0.0    5   CP    00.03.01.696  00.03.38.821        1.01     1.22      0.15       0.72   0.87
KEY4       A    50    0      2  NO     0.0    5   CP    00.03.06.431  00.03.44.484        1.04     1.25      0.15       0.74   0.89
CIM7       A    50    0      2  NO     0.0    3   CP    00.02.18.103  00.03.00.222        1.28     1.67      0.17       0.55   0.72
CIM8       A    50    0      1  NO     0.0    3   CP    00.01.57.792  00.02.35.459        1.09     1.44      0.15       0.47   0.62
CFD        A    50    0      0  NO     0.0    2   CP    00.00.24.052  00.00.31.195        0.33     0.43      0.03       0.10   0.12
CFF        A    20    0      1  NO     0.0    2   CP    00.02.07.596  00.02.16.216        1.77     1.89      0.03       0.51   0.54
*PHYSICAL*                                                            00.37.42.796                           8.98              8.98
                                                        ------------  ------------                         ------     ------ ------
  TOTAL                                                 05.50.28.671  06.36.01.290                          10.84      83.45  94.29

CF1        A    10    0     32  NO     0.0    2   ***   00.59.59.334  01.00.00.462       49.99    50.01      0.03      99.98  100.0
*PHYSICAL*                                                            00.00.03.515                           0.10              0.10
                                                        ------------  ------------                         ------     ------ ------
  TOTAL                                                 00.59.59.334  01.00.03.978                           0.13      99.98  100.1

CFE        D

RMF Monitor III CPC Capacity Report
The new Monitor III CPC Capacity Report provides the capability to monitor online software pricing related values as well as LPAR CPU values for all logical partitions in the CPC.

WLM LPAR CPU management allows pricing based on partition capacity. By a new method called 'soft cap,' WLM ensures that the average CPU consumption of a partition does not exceed a defined capacity value (in millions of unweighted CPU service units per hour - MSU/h) over a defined period of time.

WLM allows the actual workload to rise above the defined MSU limit but takes care that the 4 hour rolling average stays below. This is done by dynamically turning capping on and off. Prerequisites are uncapped LPARs with shared CPs.

With this new report you can compare the defined capacity limits against the actual MSU consumption for all partitions of the CPC. In addition, the report contains MSU information related to the last 4 hours, for the partition RMF is running in. WLM Capping % shows you how often WLM has soft capped the partition.

The CPC Capacity Report consists of two subsections:

                        RMF V1R2   CPC Capacity                   Line 1 of 21
Command ===>                                                  Scroll ===> CSR

Samples: 29      System: SYSF  Date: 12/07/01  Time: 12.54.30  Range: 30    Sec

Partition:   SYSF       2064 Model 107
CPC Capacity:     253   Weight % of Max: 62.0       4h MSU Average:    33
Image Capacity:    50   WLM Capping %:    0.0       4h MSU Maximum:    76

Partition  --- MSU ---  Cap  Proc    Logical Util %   - Physical Util % -
             Def   Act  Def   Num    Effect   Total   LPAR  Effect  Total

*CP                                                   10.7    82.4   93.1
CFD            0     0  NO    2.0       0.3     0.4    0.0     0.1    0.1
CFF            0     1  NO    2.0       1.6     1.8    0.0     0.5    0.5
CF1            0     0  NO    1.0       0.1     0.1    0.0     0.0    0.0
CIM7           0     2  NO    3.0       1.3     1.6    0.2     0.5    0.7
CIM8           0     1  NO    3.0       1.1     1.5    0.2     0.5    0.6
KEY3           0     2  NO    5.0       1.0     1.2    0.1     0.7    0.9
KEY4           0     2  NO    5.0       1.1     1.3    0.2     0.8    0.9
SYSA           0     5  NO    5.0       3.1     3.3    0.2     2.2    2.4
SYSB           0     5  NO    5.0       2.8     3.1    0.2     2.0    2.2
SYSC           0     4  NO    5.0       2.1     2.3    0.1     1.5    1.6
SYSD           0    29  NO    4.0      22.5    22.6    0.1    12.9   12.9
SYSE          45    54  NO    4.0      42.5    42.8    0.1    24.3   24.5
SYSF          50    75  NO    5.0      47.2    47.4    0.2    33.7   33.9
VMA            0     7  NO    7.0       2.8     3.1    0.3     2.8    3.1
PHYSICAL                                               8.9            8.9

*ICF                                                   0.1     100    100
CF1            0    32        1.0       100     100    0.0     100    100
PHYSICAL                                               0.1            0.1

The report header section offers MSU related values (measured in MSUs/hour) within the scope of the partition that requested the report (home partition), as well as the processor type and model, e.g. the CPU capacity available to the CPC (CPC Capacity), the CPU capacity available to the MVS image (Image Capacity) and the MSU average and maximum values during the last 4 hours. Note that the maximum value can be greater than the defined capacity for the requesting partition. The WLM Capping % value denotes the percentage of time when WLM capped the partition because the 4 hour MSU value exceeds the defined capacity limit.

The partition data subsection displays information for all partitions belonging to the CPC, grouped by partitions related to CP processors and those related to ICF and IFL processors. The partition identified by the name PHYSICAL is not a configured partition. Data reported in this line include all of the uncaptured time which was used by LPAR but could not be attributed to a specific logical partition. Besides the capacity-related fields (MSU Def/Act) you get information about the processor utilization.

RMF PM Java^TM  Client
The RMF PM Java^TM  Client has been enhanced to support the software pricing relatedmetrics.

With the installation of PTF UW99366 for z/OS 1.2 (APAR OW49807), there is a new resource called CPC under the SYSPLEX resource.

You should install APAR fixes OW51646 and OW52087. Those APARs are also interesting for older releases of OS/390 and z/OS. The related PTF numbers are UW84849 and UW85020 for OS/390 V2R6, UW84850 and UW85021 for OS/390 V2R7-V2R9, UW85851 and UW85022 for OS/390 V2R10 and z/OS V1R1, UW84852 and UW85023 for z/OS V1R2). Under the CPC resource you can find the logical partitions (LPARs) of the CPC.

As in Monitor III CPC report, LPAR CPU data was added. For example, in the LPAR DataView above, you can see metrics like % effective logical dispatch time. Please note the partitions monitored could also be z/VM or Linux for zSeries partitions.

The history of actual versus defined MSUs can be visualized by means of the Plot and Save Series dialog:

Of course, the defined MSUs are constant, but adding this series to your DataView makes it easier to compare defined MSUs against actual MSUs. You can see that the actual MSUs are above the defined MSUs for quite a long period of time. At the end, the load goes down to zero.

New Metrics in RMF PM
The following metrics are new in RMF PM

• Resource CPC:
  • actual/ defined MSU (CP/ICF/IFL) by partition
  • capacity (MSU/h)
  • % effective/ total physical utilization (CP/ICF/IFL) by partition
• Resource LPAR:
  • actual/ defined MSU
  • average number of logical processors
  • four hour MSU maximum/ average
  • image capacity (MSU/h)
  • remaining time until capping in seconds
  • % weight of max
  • % WLM capping
  • % effective/ total physical/ logical utilization
  • % total lpar management
• Resource SYSPLEX:
  • actual MSU (CP) by partition
  • four hour MSU average by partition
  • image capacity (MSU/h) by partition
  • remaining time until capping by partition
  • % WLM capping by partition
  • % weight of max by partition
  • % total physical utilization (CP/ICF/IFL) by partition/ CPC

Console Messages for WLM Capping
WLM starts capping and you want to have a console message for it? Sure, you can set up IBM System Automation for OS/390 or Tivoli NetView to react properly to it.

This technique for exception-based monitoring is not restricted to WLC related metrics. Please refer to the RMFCS chapter in the RMF User's Guide (SC33-7990) if you would like to have WTO exceptions for generic Monitor III based performance data.

Please install PTF UW85023 (APAR OW52087) to get the sample program SYS1.SERBCLS(ERBR3CPC) discussed in this section.

You can easily setup RMF for WTOs with MSU related Alerts: just use the generic RMF batch feature to direct any Monitor III report metric to the console!

SYS1.SERBCLS(ERBR3CPC) works as follows:

• It runs the Monitor III Reporter batch address space and produces the new CPC Capacity Report.
• It extracts the actual WLM Capping % and the remaining time until capping from the report table ERBCDCT3.
• It generates a WTO and diplays the actual values.

• Edit SYS1.SERBCLS(ERBR3CPC) and set your variables and thresholds. The default is the WLM Capping % variable with a threshold of zero. In other words, you will receive a WTO for all Monitor III mintimes with a capping percentage of greater than zero. Alternatively you can allocate your own data set HLQ.RMFM3B.SERBCLS and make the changes in your private copy of ERBR3CPC. In this case you should add this data set to the SYSPROC3 concatenation of the procedure RMFM3B in SYS1.PROCLIB.

• If you want to adjust the general session options (for example the Monitor III reporting range) or modify the WFEX exceptions for your batch session, it is recommended to generate a copy of your Monitor III Report Data Tables for batch: run REXX Procedure ERBM3BWX which brings up a Monitor III Reporter session and perform your adjustments. As a result, all Monitor III Data Tables are copied to the data set HLQ.RMFM3B.ISPTABLE.

• If you want to work with the Monitor III session defaults, no table copies are needed. In this case you can simply deactivate the references to HLQ.RMFM3B.ISPTABLE within SYS1.PROCLIB(RMFM3B).

• Start the RMF Monitor III batch procedure with your personal high level qualifier and the new CPC Capacity Report as selected report:

  sRMFM3B,HLQ=myUSERID,REPORT=CPC

• the actual WLM Capping percentage
• or - if capping is not in effect - the projected number of seconds when WLM is going to cap the partition.

If you want to terminate the WTO generation you can simply stop the RMFM3B address space by means of the Cancel command.

Conclusion
The enhancements in RMF Monitor i and Postprocessor allow you to do long-term monitoring of MSUs consumed by your business workloads. If you want online monitoring or problem drill-down, you can use Monitor III with its frontend RMF PM. In RMF PM, you have the metric remaining time until capping in seconds by partition. So you can anticipate capping before it actually takes place. If you want to automatically react on critical performance situations related to software pricing with your automation product of choice, you can generate WTO messages.

Your feedback is important to us
The RMF Newsletter appears on a regular basis. It is one way to keep you informed about the latest news for z/OS RMF. You also should note that the Newsletters and tools such as RMF PM, rmfpms for Linux (for S/390 & zSeries), the Spreadsheet Reporter, and more can be obtained from the RMF home page.

If you like to comment on 'RMF Newsletters' or if you are interested in other topics contact RMF Development directly. Just send a note to rmf@de.ibm.com.

Linux is a registered trademark of Linus Torvalds.