Interpreting Strobe performance profile for Adabas regions
An Adabas region is an MVS address space that processes requests to access Adabas databases from teleprocessing regions, such as CICS and TSO, and batch processing programs. Requests to move data from and to the caller’s address space are made through the Adabas SVC.
When Strobe measures an Adabas region, the Strobe Performance Profile for that region shows the following:
- CPU activity in Adabas modules and in user exit routines
- Function descriptors for the Adabas routines
- CPU usage attributed to callers within the user regions that caused the activity, identified for the following:
- Batch job, by the job name
- TSO region, by the TSO user ID
- CICS region, by the logical terminal name.
When Strobe measures a 3GL program, the Strobe Performance Profile reports CPU wait time for the ADALNK module back to the 3GL (Assembler, PL/I, FORTRAN, or COBOL) program that called it.
This section describes a Performance Profile of an Adabas region and provides examples of the key reports.
Measurement Session Data Report
The Measurement Session Data report describes the environment during a measurement session. This example (See the following figure) shows a busy Adabas region. The SUBSYSTEM field of the report shows the version number of Adabas in the measured region. One or more CPUs were active during 8% of the 5-minute measurement session. The Adabas databases were also relatively busy, averaging more than 32 accesses per second.
Measurement Session Data Report for Adabas Regions
Time Distribution and Resource Demand Distribution Reports
The Time Distribution of Activity Level report (Time Distribution of Activity Level Report for Adabas Regions) and the Resource Demand Distribution report (Resource Demand Distribution Report for Adabas Regions) show the level of all resource utilization in the Adabas region during the measurement session. The reports show the execution of two Adabas tasks and the I/O activity of the Adabas data sets. These data sets include the databases and the associator, work, and log data sets. In this example, the associator data set, DDASSOR1, which contains access information for the databases, and DDDATAR1, the database data set, are the most heavily accessed.
Time Distribution of Activity Level Report for Adabas Regions
The I/O activity for each of the data sets in the Adabas region seldom reaches levels as high as 80 or 90 percent. This lack of high I/O activity indicates that you can probably improve performance by increasing the overlap of data set access. The Resource Demand Distribution report indicates that the Adabas data sets DDDATAR1 and DDASSOR1 incur wait time of around 25% of total run time and show high solo execution time percentages. These values also hint that the lack of data set overlap could be a problem. The I/O Facility Utilization Summary report for Adabas Regions (I/O Facility Utilization Summary Report for Adabas Regions) indicates where these data sets are located and displays their I/O access. This report can show you whether performance can be improved by increasing data set overlap.
Resource Demand Distribution Report for Adabas Regions
Working Set Size Through Time Report
The Working Set Size Through Time report (See the following figure) shows a fairly uniform working set size throughout the measurement session. The Measurement Session Data report (Measurement Session Data Report for Adabas Regions) confirms a relatively low paging rate of 1.30 pages per second.
Working Set Size Through Time Report
Wait Time by Module Report
The Wait Time by Module report (See the following figure) shows the majority of wait time in the Adabas nucleus module that waits for the completion of all events. This wait time includes I/O activity and time spent waiting to work. To see wait time caused by the Adabas data set access, refer to the Resource Demand Distribution report for Adabas Regions (Resource Demand Distribution Report for Adabas Regions).
Wait Time by Module Report for Adabas Regions
I/O Reports
Three different types of reports provide information about I/O and data set access.
Data Set Characteristics Report
The Data Set Characteristics report (See the following figure) shows, for each data set accessed during the measurement session, the access method (Adabas data sets are BDAM-based), block size, and data set name. The number of I/O operations performed (EXCPs) during the measurement session appears as well. In the example shown, the EXCP counts confirm the heavy use of the associator data set, DDASSOR1, and the database DDDATAR1.
Data Set Characteristics Report
DASD Usage by Cylinder Report
A DASD Usage by Cylinder report (See the following figure) shows data set access by unit, volume, ddname, and cylinder.
In this example, a small portion of the database accounts for most of the DASD usage. Identifying the frequently used data may suggest opportunities for improvement.
For the Performance Profile, this report was compressed by the specification of the parameter DASD=.1 as a report option. As a result, Strobe combined activity into a single line for contiguous cylinders that show less than 0.1 percent activity.
DASD Usage by Cylinder Report
I/O Facility Utilization Summary Report
The I/O Facility Utilization Summary report for Adabas Regions (I/O Facility Utilization Summary Report for Adabas Regions) summarizes I/O access by device, volume, and ddname. In this example, the database, the associator data set, and the log data set are on the same unit. I/O overlap should increase here if the database and associator data set were assigned to separate units. This reassignment should result in better response time in the online regions and faster batch processing jobs, as suggested by the Resource Demand Distribution report for Adabas Regions (Resource Demand Distribution Report for Adabas Regions).
This example also shows that unit 31A is 51% utilized, implying that there must be significant queuing on this device.
I/O Facility Utilization Summary Report for Adabas Regions
Reports Showing CPU Activity
Several Strobe reports provide information about CPU usage by and within a program.
Most Intensively Executed Procedures Report
The Most Intensively Executed Procedures report (See the following figure) shows the ten most frequently used programs in the Adabas region. The Adabas nucleus modules ADANC6, ADANC2, and ADANC8 are the most heavily executed.
Most Intensively Executed Procedures Report
Program Section Usage Summary Report
The Program Section Usage Summary report for Adabas Regions (See the following figure) shows the distribution of CPU time used by each active control section of each active module in the Adabas region. Adabas and IBM system modules are compressed and shown as pseudo-sections under the pseudo-module .SYSTEM.
Most of the CPU time is spent in Adabas modules. In this example however, some of the time is spent in a user exit module, UE4MAINM. Review the Program Usage by Procedure reports for these control sections to locate the active areas of code and identify any performance improvement opportunities.
Program Section Usage Summary Report for Adabas Regions
Program Usage by Procedure Report
The Program Usage by Procedure report for Adabas Regions (See the following figure) shows, for IBM and Adabas system routines, the modules and control sections responsible for the CPU activity in the Adabas region. Function descriptors appear for each control section or, if Strobe has not gathered control section information, for each module.
Program Usage by Procedure Report for Adabas Regions
Reports Showing CPU Usage Attributed to User Regions
Strobe provides Performance Profile reports that can trace CPU usage to user regions.
Callers’ DB Usage Summary Report
The Callers’ DB Usage Summary report (Callers’ DB Usage Summary Report) shows the percentage of total CPU time consumed by each caller of Adabas services.
Callers are identified for:
- Batch job, by the job name
- TSO region, by the TSO user ID
- CICS region, by the logical terminal name.
Overhead activity in an Adabas region is attributed to the pseudo-section .ADABAS.
When a task other than an Adabas system service is executing, Strobe for Adabas/Natural attributes the CPU time to a pseudo-section whose name begins with a period (.) followed by the first seven characters of the module name in which execution of the task was initiated.
During the measurement session, this Adabas region was accessed by one batch job, WDC9582, and four CICS terminals. The batch job was the heaviest consumer of CPU time. If the CICS response time suffers because of contention for database access, the batch program might be scheduled to run when the CICS region is less busy or has been shut down.
Callers’ DB Usage Summary Report
Callers’ DB Usage by Control Section Report
A Callers’ DB Usage by Control Section report (See the following figure) is produced for each of the database users identified in the Callers’ DB Usage Summary report. The report header line identifies the calling region. Each detail line shows solo and total CPU time spent on behalf of the caller in either an Adabas or IBM system module or in a user-written control section.
Callers’ DB Usage by Control Section Report
