| |
|---|
|
|---|
| Name of policy used by the service definition |
| Date and time service policy was activated |
| Date and time this policy was installed |
| User ID of person who installed the policy |
|
|---|
| Name and description of the workload in the service class definition in the workload manager |
SERVICE CLASS, REPORT CLASS | Name and description of the service class or report class in the service definition |
| Period number of the service or report class |
| Importance level of the performance goal, ranging from 1 (highest) to 5 (lowest) This field is not present for report classes and not applicable to service classes having system or discretionary goal. |
| Name of the resource group to which the service class is assigned Resource groups are used to set a minimum or limit the amount of CPU capacity available to one or more service classes. This field is not present for report classes. |
| An indication that I/O priority groups are enabled in the Workload Manager (WLM) and the workload is a high priority If the field does not appear, either I/O priority groups have not been enabled or the workload is not a high priority. |
| Indicates that a service class is not eligible for honor priority processing. When HONOR PRIORITY=NO, specialty engine eligible work in this service class will not be offloaded to CPs. |
HOMOGENEOUS REPORT PERIOD | A report class period in which transactions from a single service class contribute data to it; the report also shows the service class |
HETEROGEOUS REPORT PERIOD | A report class period in which transactions from more than one service class contribute data to it |
SERVICE UNIT COEFFICIENTS | Numbers by which I/O, TCB, SRB, and storage service units are multiplied |
| Whether WLM dynamically adjusts the priority of non paging DASD I/O requests to meet goals |
| zAAP and zIIP Normalization Factors: used to convert zAAP and zIIP processor usage to equivalent CP usage - CPusage = zAAP usage * zAAP Normalization Factor
- CPusage = zIIP usage * zIIP Normalization Factor
These formulas allow for the possible differences in processor speed between CPs and zAAPs or zIIPs. |
| Column heading of transaction counts Possible values are: - AVG—average number of active transactions (not necessarily in central storage)
- MPL—average number of transactions resident in storage
- ENDED—total number of transactions completed in the report interval
- END/SEC—average number of transactions completed per second
- # SWAPS—number of swap sequences during the interval
- EXECUTD—number of execution phases completed as reported by subsystem work managers
- AVG ENC—average number of independent enclaves
- REM ENC—average number of foreign enclaves imported from a remote system
- MS ENC—average number of multisystem enclaves exported to and executing on a remote system
|
TRANS. TIME HHH.MM.SS.TTT | Column heading of average transaction times - ACTUAL—average transaction response time
- EXECUTION—average transaction execution time
- QUEUED—average queue delay time
- For batch jobs, it is the time jobs spent waiting for an initiator.
- For TSO users, it is a portion of the LOGON process.
- For APPC, it is the time an APPC request spent on an APPC queue.
- R/S AFFINITY—average time that batch jobs were ineligible to run because the resources the job had affinity to was unavailable
- INELIGIBLE—average time that batch jobs spent on job queues (after JCL conversion) while ineligible to run for reasons other than resource affinity, including operator hold, delays due to duplicate job names, delays due to job class limits
- CONVERSION—average time that batch jobs spent on JCL conversion
- STD DEV—measure of how widely transaction response times varied from the average value
|
| Column heading of non-paging DASD I/O requests - RATE—average number of non-paging DASD I/O requests completed per second
- RESP—average response time of non-paging DASD I/O requests in milliseconds
- CONN—average connect time in milliseconds, including data transfer and search time
- DISC—average disconnect time in milliseconds
- PEND—average pending time in milliseconds, including channel & control unit contention, device busy on another system
- IOSQ—average IOS queue time in milliseconds, that is, time waiting for a device while it is busy performing I/O
|
| Column heading of the number of service units of each type consumed in the interval - CPU—number of CPU service units that were used by transactions while executing in TCB mode or preemptible SRB mode
- SRB—number of CPU service units that were used by transactions while executing in non-preemptible SRBs
- I/O—number of I/O service units used
- MEM—number of memory service units used
- TOTAL—total number of I/O, TCB, MEM, and SRB service units used
- SU/SEC—rate at which service units were used per second
- ABSRPTN—service units used per second while transactions were resident in central storage
- TRX SRV—service units used per second while transactions were active (not necessarily in central storage)
|
| Column heading for CPU seconds that services consumed during the reporting interval; values shown are in seconds - CPU—number of CPU seconds that were used by transactions while executing in TCB mode or preemptible SRB mode
- This value reflects CPU time spent on all processor types.
- SRB—number of CPU seconds that were used by transactions while executing in non-preemptible SRBs
- RCT—region control task time in seconds
- I/O INT—I/O interrupt processing time in seconds
- HS SERV—hiperspace service time in seconds
- ZAAPonCP—service time for work that was executed on standard CPs eligible to run on zAAPs; value is in seconds and is normalized to the equivalent time on a standard CP
Note zAAPs do not necessarily run at the same speed as standard CPs. To provide a proper basis of comparison, the zAAP time values are normalized to represent the amount of time the task would have taken if executed on a standard CP. 1 - ZAAP—zAAP time in seconds 2
- ZIIPonCP—service time for work that was executed on standard CPs eligible to run on zIIPs; value is in seconds 1
- ZIIP—zIIP time in seconds 3
|
| CPU time, in seconds, that transactions in this group were running at a promoted dispatching priority The reasons for promoting the work unit is to: - BLKQL—relieve a lower dispatching priority blocked workload
- ENQUE— relieve an enqueue conflict
- RSRCE— relieve a resource contention
- LOCK—shorten the lock hold time of a local suspend lock held by the work unit
- SPVSR—CPU time consumed by work units while promoted by the z/OS supervisor.
|
| Column heading for percentage of the reporting interval that transactions used standard CPs, zAAPs, or zIIPs Note It is possible for values shown in this column to exceed 100% when an LPAR has more than one logical processor or input contains data from more than one system. When SMT=2 (or higher), the APPL% is further divided by the multi-threading maximum capacity factor. - CP%—percentage of the report interval that transactions used standard CPs; the sum of CPU, SRB, RCT, I/O INT, and HS SERV service seconds, minus normalized zIIP and zAAP service seconds, divided by the report interval
- ZAAPonCP%—percentage of the report interval that service time was executed on standard CPs but was eligible to run on zAAPs 1
- ZAAP%—percentage of the report interval that zAAP time was used 2
- ZIIPonCP%—percentage of the report interval that service time was executed on standard CPs but was eligible to run on zIIPs 1
- ZIIP%—percentage of the report interval that zIIP time was used 3
|
| Column heading for storage frames allocated to address spaces - AVG—average central and expanded storage, in frames, allocated to an address space while resident in storage; it is page residency time divided by address space residency time
- TOTAL—average central and expanded storage, in frames, allocated to all address spaces during the report interval; it is page residency time divided by report interval
- CENTRAL—average central storage, in frames, allocated to all address spaces during the report interval
- SHARED—average shared storage, in frames, allocated to all address spaces during the report interval
|
| Column heading for paging, in pages per second - SINGLE—rate at which individual pages were transferred due to demand paging
- BLOCK—rate of pages transferred (paged in) from auxiliary storage to central storage in blocks, as part of swap-in or directed page ins
- SHARED—rate of shared pages transferred (paged in) from auxiliary storage to central storage
- HSP—rate per second of hiperspace pages transferred (paged in) from auxiliary storage to central storage while transactions were active
- HSP MISS—rate of expanded storage only (ESO) hiperspace read misses
A miss occurs when a read to expanded storage is issued and the requested page resides in auxiliary storage.
|
DISCRETIONARY GOAL MANAGEMENT | Whether a discretionary goal is assigned to the service class and tenant report class periods that are connected to the resource group and are therefore excluded from workload management Possible values are YES and NO. |
| - TOTAL—percentage of each type of processor time used by all transactions
- CP—total percentage of the general purpose processor time used by all transactions
- ZAAP/ZIIPonCP— total percentage of the processor time used by all transactions that are eligible to run on a specialty processor, but were executed on a general purpose processor
- ZAAP/ZIIP— total percentage of the specialty processor time used by all transactions
- MOBILE—percentage of each type of processor time that is used by transactions that are classified with the MOBILE reporting attribute
- CP—total percentage of the general purpose processor time used by transactions that are classified with the MOBILE reporting attribute
- ZAAP/ZIIPonCP—total percentage of the processor time used by transactions that are classified with the MOBILE reporting attribute, and are eligible to run on a specialty processor, but were executed on a general purpose processor
- ZAAP/ZIIP—total percentage of the specialty processor time used by transactions that are classified with the MOBILE reporting attribute
|
SERVICE CLASSES SERVED LINE |
|---|
| If the address spaces in the reported service class performed work for other service classes, this field shows how many times each other service class was served NONE means no other service classes were served, and N/A appears for Report Classes, Report Class Summaries, Workload Summaries, and Service Policy Summaries. |
|
|---|
| Line describing the goal of the service class; there are five types of goals: - System—Unique to SYSTEM service classes, the system goal always has a dispatching priority of 255 and an I/O priority of 255. This type of goal appears on the report as GOAL: SYSTEM.
- Average Response Time—The average response time is the expected amount of time required to complete the work submitted under the service class, in milliseconds, seconds, minutes, and hours. It includes only the time spent on z/OS; for example, it will not include network time. This type of goal appears on the report as GOAL: AVERAGE RESPONSE TIME OF nn.nn.nn.nnn( HH.MM.SS.TTT).
- Percentile Response Time—The percentile response time is the percentage of work in that period that should be completed within the response time (for example, 75 percent of transactions ended in 0.6 of a second). This type of goal appears on the report as GOAL: nn.n PERCENTILE OF RESPONSE TIME WITHIN nn.nn.nn.nnn( HH.MM.SS.TTT)..
- Execution Velocity—Execution velocity defines how fast work should run when ready, without being delayed for processor, storage, I/O access, and queue delay. Execution velocity goals are intended for work for which response time goals are not appropriate, such as started tasks, or long running batch work. This type of goal appears on the report as GOAL:EXECUTION VELOCITY: nnn.n.
- Discretionary—Work that is described as discretionary is work that should run only when service classes with a higher importance do not need the resources to achieve their respective goals. This type of goal appears on the report as GOAL:DISCRETIONARY.
|
EXECUTION DELAY/RESPONSE TIME SUBSECTION |
|---|
EXECUTION VELOCITY MIGRATION | Helps you plan your choice of velocity goals The next two fields relate to Execution Velocity Migration. |
| I/O Priority Management; a WLM facility that gives you the flexibility to control nonpaging DASD I/O requests that are queued because the device is busy This facility can be set to On or Off in the WLM ISPF application by using the Service Coefficient/Service Definition Options panel; the default is Off. - If set to On: WLM dynamically adjusts the I/O priority based on goals and I/O activity and includes the I/O information in calculating execution velocity. (For more information, see the IBM z/OS MVS Planning Workload Management manual.)
- If set to Off: The number of samples of work using nonpaging DASD I/O resources and the number of samples that are delayed for nonpaging DASD I/O are excluded from the execution velocity calculation.
The value in I/O MGMT is the achievable execution velocity, including nonpaging DASD I/O resources and delay samples (that is, the execution velocity if I/O Priority Management is set to On). After this line, you see the actual overall execution velocity of the same service class. If I/O Priority Management is enabled in the service definition, the value in I/O MGMT should match the execution velocity in the overall value (field EX VEL in the report). If your organization has not set I/O Priority Management to On, you can use the I/O MGMT value as a guide in your execution velocity planning exercise. |
| Initiator Management; the value of achievable execution velocity if batch initiator delay samples were included in the velocity calculation Similar to I/O MGMT, this number helps you if you want to know what the execution velocity will be if batch initiator management is enabled. If you have Initiator Management enabled, the value in INIT MGMT will match the value in EX VEL in the report. |
| For average response time goal, the expected response time is formatted on the GOAL line and the actual response time is on the subsequent lines for sysplex and individual systems For response time with percentile goal, the expected percentage of transactions to be completed within the specified response time is formatted on the GOAL line and the actual percentage is on the subsequent lines for sysplex and individual systems. For system or discretionary goal, SYSTEM or DISCRETIONARY is printed. This column is not applicable to report classes. |
| For velocity goal, the expected execution velocity is formatted on the GOAL line The actual execution velocity is formatted on the subsequent lines for sysplex and individual systems, regardless of goal type. |
| Performance index Note This column is not applicable to report classes and service classes of system or discretionary goal. |
| Average number of address spaces and enclaves that contributed delay and using samples |
| - CPU—ratio of samples that are using standard CPs compared with all using and delay samples
- ZAAP—ratio of samples using zAAPs compared with all using and delay samples
- ZIIP—ratio of samples using zIIPs compared with all using and delay samples
- I/O—ratio of nonpaging DASD I/O using samples compared with all using and delay samples
|
| TOTAL—percentage of all delay samples used in the calculation of execution velocity This field does not necessarily equal the sum of delay reason percentages listed (for example, if I/O priority management option is inactive, I/O delay reason is not included.) At most, five delay reasons are listed in descending order at the sysplex level: - CPU (CPU delay)—work has been delayed while either waiting to be dispatched on a standard CP or waiting for the local lock
- CPU CAPP (CPU capping delay)—work is nondispatchable because a resource group maximum is being enforced
- DASD (nonpaging DASD I/O)—This delay is not included in TOTAL if WLM does not manage I/O priority.
- MPL (MPL delay)—work is delayed for storage due to multiprogramming level constraint
- PGIN COMM—page in for common storage
- PGIN EHSP—page in for ESO hiperspaces
- PGIN HSP—page in for standard hiperspaces
- PGIN PRIV—page in for private storage
- PGIN VIO—page in for VIO storage
- PGIN XMEM—cross memory page in
- SERV HSP—server hiperspace page in
- SERV MPL—server MPL delay
- SERV PRIV—server private area page in
- SERV QUE—server queue delay—work is waiting for a server
- SERV SWIN—server swap-in
- SERV VIO—server VIO page in
- SWAP IN—swap-in delay
- ZAAP (zAAP delay)—work that has been delayed while waiting to be dispatched on a zAAP
- ZIIP (zIIP delay)—work that has been delayed while waiting to be dispatched on a zIIP
|
| Work is delayed, but none of the reasons listed under EXECUTION DELAYS % above apply |
| Work is in STIMER wait, TSO terminal wait, APPC wait, or is an initiator waiting for work |
| Work in this service class has been reset by the RESET jobname, QUIESCE command, which swaps out a swappable address space or gives the lowest possible performance characteristics to a nonswappable address space |
| Crypto using state—a TCB or SRB was found to be using a cryptographic asynchronous message processor (CAP) or an adjunct processor (AP) |
| Crypto delay state—a TCB or SRB was found to be waiting for a cryptographic asynchronous message processor (CAP), an adjunct processor (AP) or a processor feature queue |
| Percentage of samples where work is holding a resource in contention, as reported by a resource manager by way of the IWMCNTN service |
| Percentage of samples where work is waiting for a resource in contention, as reported by a resource manager by way of the IWMCNTN service |
RESOURCE MANAGER STATES SUBSECTION |
|---|
| Subsystem type, as specified in the classification rules |
| States sampled in the begin-to-end phase of a transaction (BTE) or in the execution phase (EXE) |
| Percentage of the response time that a transaction spends in the begin-to-end or execution phase Note Long-running and never-ending transactions that were not completed during the report interval contributed state samples but not response time, causing this transaction to appear inflated, perhaps even registering above 100%. |
RESOURCE MANAGER STATES % | Ratio of the specified state samples compared with total state samples - ACTIVE—a program was executing on behalf of the work request, from the perspective of the work manager; does not mean that the program is active from the MVS perspective
- SUB—program was the subsystem itself
- APPL—program was an application invoked by the subsystem
- READY—a program was ready to execute on behalf of the work request, but the work manager has given priority to another work request
- IDLE—work manager saw a transaction as idle
- DELAYS TOTAL—percentage of all delay state samples. At most, six delay reasons are listed in descending order at the sysplex level:
- BPMI—buffer pool miss
- CONV—waiting for conversation
- DIST—waiting for distributed request; some function or data had to be routed prior to resumption of the work request
- I/O—waiting for I/O
- LTCH—waiting for new latch
- LOCK—waiting for lock
- LSES—waiting for a session to be established locally on the current MVS image
- NSES—waiting for a session to be established somewhere in the network
- PROD—waiting for another product
- SSES—waiting for a session to be established somewhere in the sysplex
- TIMR—waiting for a timer
- UNKN—waiting for an unidentified resource
- SSLT—waiting for an SSL thread
- REGT—waiting for a regular thread
- WORK—waiting for a registration worktable
|
RESOURCE MANAGER STATES % | CONTINUATION Percentage of logical continuation state samples: - LOC—logical continuation on the local system; subsystem work managers function shipped a transaction to another component within the same system
- REM—logical continuation on a remote system in the sysplex; subsystem work managers function shipped a transaction to another component on another system
- SPLX —logical continuation somewhere within the network; subsystem work managers function shipped a transaction to another component within the network
|
TRANSACTION RESPONSE TIME DISTRIBUTION SUBSECTION |
|---|
TRANSACTION RESPONSE TIME DISTRIBUTION | For response time with percentile goal, this table shows the distribution of completed transactions |
| Percentage of goal reached for a specific column |
| Response times of response time buckets; unit of measure is determined dynamically The unit can be milliseconds (MSEC), seconds (SECS), minutes (MINS), or hours (HRS). |
| Percentage of transactions completed in the reporting period represented by a response time bucket |
| Percentage of transactions completed with a response time less than or equal to the value displayed in the RESP TIME line |
| Number of transactions completed in the reporting period represented by a response time bucket |
| Number of transactions completed with a response time less than or equal to the value displayed in the RESP TIME line |
EXECUTION VELOCITY RESPONSE TIME DISTRIBUTION SUBSECTION |
|---|
| For Execution Velocity goals, this table shows the distribution of completed transactions. Note One such table is produced for each system. |
| The table heading consists of one line showing the following: - SYSID: the SYSID of the system being reported in the table
- Average Response Time (HH:MM:SS.TTT): The midpoint of all response times measured during the period
- # of Changes: Number of midpoint changes that occurred during the interval
- Time of Last Change: latest point in time when a midpoint change occurred
|
| The percentage of the goal that is reached for each column |
| Execution Velocity of each column |
| The percentage of the whole for each column |
| The accumulated percentage of this column and all columns to the left of it |
| Number of transactions in this column |
| The accumulation of transactions in this column and all columns to the left of it |
SYSTEMS SUMMARY SUBSECTION |
|---|
| System ID of the systems contributing data to this report |
| Start date and time of the earliest record |
| End date and time of the most recent record |
| Total recording intervals of records produced on the systems in ddd.hh.mm.ss format |
| z/OS release in the ZVvvrrmm format for z/OS release vv.rr.mm |
| Suffix of the IEAOPTxx parmlib member containing the parameters used by SRM |
| Unweighted CPU service units per second of TCB or SRB execution time, dependent on processor model |
| Effective CPU capacity available percentage Note When in the range 1 to 99, machine is operating at reduced capacity. When at 100, the machine is operating at full capacity. 'N/A' is displayed, if this field cannot be calculated. If the machine is operating at a reduced capacity, the reason appears on the CPU Utilization Report in the 'CHANGE REASON' field of the Summary Section. |
BOOST (PTF BQM1631 applied) | Boost class and type that was active for the interval Possible values for boost type: - Z—zIIP Boost
- S—Speed Boost
- B—Both zIIP Boost and Speed Boost
- N —None of the above boost types
Possible values for boost class: - SHDN—Shut Down Boost
- IPL—IPL Boost
- NONE—None of the above classes, or no boost type is active right now
|
|
|---|
RESOURCE GROUP/ SERVICE CLASS | NAME—resource group and service classes that are associated with the resource group DESCRIPTION—resource group and service class descriptions |
| |
| Service units expressed three ways, with a reporting scope of: the resource Group, an MVS system in the sysplex, a service class or tenant report class belonging to the resource group - #CPS—CPU consumption in number of general purpose processors
- MSU—CPU consumption in MSU
- SU/SEC—CPU consumption in unweighted CPU and SRB service units per second
|
| - MIN—minimum capacity of the resource group
- MAX—maximum capacity of the resource group
- DEFINED AS—method by which the resource group capacity is defined
Possible values are as follows:- SERVICE UNITS/SEC—service units per second
- PCT OF LPAR SHARE—percentage of partition share
- PCT OF SINGLE PROC—percentage of a single processor
|
| Whether the resource group is running on a sysplex or system |
| Amount of memory used by the resource group |
| Resource group memory limit for each system in the sysplex the memory consumption of work running in the address space that is associated with the resource group is limited to this value. |