gtpm1m3tTPF V4R1 Migration Guide: 3.1 to 4.1

Data Collection and Reduction, and Performance Monitoring

The performance of the TPF 4.1 system must be monitored to organize system resources properly for peak operating efficiency. The TPF 4.1 system provides data collection and reduction programs to measure system performance. These programs provide operational data about activities such as:

With this information and the transaction history, you can determine how efficiently the TPF 4.1 system is running, where bottlenecks occur, and what changes can improve system performance.

The TPF 4.1 system is enhanced to provide statistical data about each ECB. To aid performance monitoring and capacity planning, this data includes:

The preceding data is available when the ECB exits through the EXI user exit.

In addition, intercept hooks were added for selective collection of statistics on Enter/Back, VFA, and SON macros. To better control the amount of data collected the DATACO (SIP) macro now has a SKIP parameter that specifies the number of entries between samplings (the default is 99). The ZMEAS command no longer supports the T parameter.

The data collection and reduction programs were rewritten substantially. Consequently, any modification that you have made to these programs will require additional effort. In addition, data collection no longer references the JDCD tape. Rather, it now uses a real-time tape named RTC.

Data collection and reduction provides:

The new blocks and structures for the TPF 4.1 system require a variety of changes in the reports:

The new data collection hooks allow a variety of macros and messages to be individually tracked. These include:

Collection of this information allows a finer focus on the activities of an individual ECB. In addition new fields were added describing CPU and file usage for each ECB. The dispatch and end times are recorded, as are the number of finds and files, and the number of primary memory requests (GETFCs). A field allows a user-specified identifier to be associated with the ECB.

See the TPF System Performance and Measurement Reference for more information about data collection and performance monitoring.

Data Collection

Data collection can be run in continuous mode or sampling mode, allowing multiple types of data to be captured while avoiding significant interference with message processing. All data collection programs write the capture data to an online tape. No attempt is made to analyze the data online, as this would have a negative impact on the system that is being measured.

The three basic techniques used for collecting data are:

Data Reduction

All data reduction is performed on an IBM Multiple Virtual Storage (IBM MVS) system. The data reduction reports are intended for use by an analyst familiar with the TPF 4.1 system. Frequency distribution reports including means, standard deviations, and variances of many parameters are available.

The aim of the initial analysis phase of a working system is to establish the normal limits for each of the key factors affecting performance. Once these limits are set and agreed to be realistic, a periodic system check becomes routine.

The analysis of performance data must always start with summary reports. These reports provide key data required for history and trend analysis. When investigating a problem area, the more detailed plot reports or the specialized reports of the DASD and message reduction programs are used. The plot reports, which show the value of each parameter sample in chronological order, are very effective for analyzing the cause-and-effect relationship between parameters.

Data Reduction Input Tape Changes

In the TPF 4.1 system, the input tape processed by data reduction changed significantly. The changes follow here.

Area of Change
Description of Change

 Input Tape 
Formally a general use tape called JCD, it is now a real-time tape called RTC.

 BS Record 
This record is no longer supported because the data was moved to the new MM record.

 DB/DE Record 
The storage commodity data was updated; the Frame/Heap usage histograms were added; the CRPA and extra PAT slots for E-type loader data were added; the size of virtual file access (VFA) was added; the ZSTRC settings were added.

 DR Record 
The resource vector table (RVT) information record was updated to contain the resource identifier (RID) since the RVT is no longer directly indexed by RID.

 DU Record 
Data collection no longer controls tape blocking, so this record is no longer supported.

 FC Record 
The program event recording (PER) device tape queue length was added.

 FF Record 
The size was increased to a 4K block and the entries are no longer sequential (unless the Skip factor is set to zero).

 FM/FN Record 
A time stamp and the program event recording (PER) device I/O response time was added.

 FQ Record 
The file queue record was updated to use fullword counters for the device queue length.

 FV Record 
A time stamp and the redefinition of the current counters.

 IS/IR/IT Record 
The DASD IPC records were deleted since DASD IPC is no longer supported.

 MI/MO Record 
This record is no longer supported because the data was moved to the new MM record.

 MM Record 
The new message intercept record contains entries that look like the old MI/MO/NI/NO/BS message intercept records.

 MxX/MxR/MxT Record 
The secondary ID of H was dropped since airlines line control (ALC) is no longer supported. The secondary ID of B was updated to contain only one bisynchronous station for each record. The secondary ID of L was updated to contain only one synchronous link control (SLC) link for each record.

 NI/NO Record 
This record is no longer supported because the data was moved to the new MM record.

 NS/NR/NT 
The SNA data record header was changed to use a fullword time stamp.

 PS/PI/PT Record 
This record is no longer supported because the data was moved to the new PP record.

 PP Record 
The old PS/PI/PT records were moved into the program intercept record. The program name is now included in the data. The origin of the data is the program allocation table (PAT) and the data is no longer sequential unless the Skip factor is set to zero.

 PX Record 
This record is no longer supported because the data is no longer available.

 SS/SR/ST Record 
The storage commodity data was updated. The IPTE and PTLB counts were added.

 SU Record 
The subsystem data record was updated to remove unused fields from the header.

 TT Record 
A new record was added to bracket the data collection period. The TT record contains the current control information that is used to run data collection.

Data Reduction Output Report Changes

The following tables show the changes to the output reports for data reduction, and detail the impact of data collection and reduction on each report in the TPF 4.1 system.

These output reports are:

General Reduction Reports

Table 48. General Reduction Report Changes

General Reduction Report Changes
Data Collection and Reduction Environment Summary Report This report is now named and the input tape name is now listed. The ZSTRC settings for the online system execution modes are listed and the online intercept Skip factor is listed.
Application Summary Report A total line was added.
Pushbutton Application Summary Report There are no changes.

System Reduction Reports

Table 49. System Reduction Report Changes

System Reduction Report Changes
System Section of the System Summary Report The IPTE and PTLB counts were added.
I-Stream Section of the System Summary Report There are no changes.
Core Pool Section of the System Summary Report This report was converted to report on the physical block types that are defined in the TPF 4.1 system (for example, LIOCB, LECB, LSWB, LFRAME, and LCOMMON).
Shutdown Section of the System Summary Report This report was converted to report shutdown conditions based on the physical block types defined in the TPF 4.1 system.
Frame Usage Histogram Report This is a new report that produces a histogram of the frames held at ECB exit.

Area
Description

 Purpose 
Reports the number of frames held at ECB exit time in histogram format.

 Value 
Used to evaluate the frame usage of the ECB.

 Option 
The SYSTEM option generates this report.
Heap Frame Usage Histogram Report This is a new report that produces a histogram of the Heap frames held at ECB exit.

Area
Description

 Purpose 
Reports the number of Heap frames held at ECB exit time in histogram format.

 Value 
Used to evaluate the the Heap (ECB private storage above 16 MB).

 Option 
The SYSTEM option generates this report.
System Plot Reports There are no changes.
System Distribution Reports There are no changes.
Interprocessor Communication Summary Report No longer supported. The data previously contained in this report is now found in the Interprocessor Communication MPIF Summary Report.
System Pools Summary Report There are no changes.
MPIF Configuration Summary Report There are no changes.
Interprocessor Communication MPIF Summary Report There are no changes.
MPIF Path Activity Report There are no changes.

File Reduction Reports

Table 50. File Reduction Report Changes

File Reduction Report Changes
Subsystem Summary Report There are no changes.
File Access Summary Report The input/output (I/O) response times were added.
VFA Summary Report In the TPF 4.1 system, the virtual file access (VFA) summary data was moved from the Random File Access Summary Report and reformatted to look like the ZVFAC IND command output. In addition, the VFA hit rate was added.

Area
Description

 Purpose 
Reports on VFA activity.

 Value 
Used to evaluate the VFA activity in the online system.

 Option 
The FILE option generates this report.
Random File Access Summary Report The virtual file access (VFA) summary data was moved to the new VFA Summary Report and the input/output (I/O) response time was added.
File Plot Reports There are no changes.
File Distribution Reports There are no changes.
Channel Path ID Status Report There are no changes.
Channel Path Activity Summary Report There are no changes.
SDA Activity Summary Report There are no changes.
Sequential File Access Summary Report Change - The mean tape queue size will be added.
File Accesses Per Record ID Report The virtual file access (VFA) fast path accesses were added as a column in this report when the device type is VFA.
File Accesses Per Cylinder Report There are no changes.
File Comparison Plot Report There are no changes.
3880 Cache Detail Report There are no changes.
3990 Cache Detail Report There are no changes.
3990 Cache Device Detail Report There are no changes.
3990 Cache Summary Report There are no changes.
3990 Device Summary Report There are no changes.

Program Reduction Reports

Table 51. Program Reduction Report Changes

Program Reduction Report Changes
Program Names and Enters Report There are no changes.
Program Details Report The columns indicating the nesting level were deleted since that data is no longer available.
Cumulative Program Enters Report There are no changes.
Cumulative Program On-File Enters Report There are no changes.
Program-Category Statistics Report There are no changes.
Package Details Report The columns indicating the nesting level were deleted since that data is no longer available.
Cumulative Package Enters Report There are no changes.
Cumulative Package On-File Enters Report There are no changes.
Subsystem Details Report This report is no longer supported. There is no replacement.
Cumulative Subsystem Enters Report This report is no longer supported. There is no replacement.
Cumulative Subsystem On-File Enters Report This report is no longer supported. There is no replacement.
Program Subsystem Counts Spread Report This report is no longer supported. There is no replacement.
Program Subsystem Rates Spread Report This report is no longer supported. There is no replacement.
Program Subsystem Percentages Spread Report This report is no longer supported. There is no replacement.
Program Categories Report References to coreslow and corefast were changed to core resident. Categories that no longer exist were removed.
Program Legend Report References to the columns for nesting levels were removed.
Weighted Message Rate Report There are no changes.

Message Reduction Reports

Table 52. Message Reduction Report Changes

Message Reduction Report Changes
Message Summary Report There are no changes.
Action Code Summary Report There are no changes.
Application Detail Report There are no changes.
City Summary Report There are no changes.
Terminal Activity Report There are no changes.
Logical Unit Activity Report There are no changes.
Message Stream Report There are no changes.