Communication subsystem contention in DB/DC and DCCTL environments

Because of the great variety of transmission devices and types of lines, each suspected problem in the communication subsystem area must be separately examined.

As described in Considerations for the communications network in DB/DC and DCCTL environments, many possible causes of delay exist within the network hardware and software. A variety of different tools might be required to investigate any suspected problems in this area.

Dynamic monitoring

It is not possible to dynamically monitor the performance of the telecommunication subsystem.

Daily monitoring

If the response time breakdown data indicates large and variable IMS output queue times, a bottleneck might exist in the IMS output message handling or on a line. Because it is not normally possible to reconfigure the TP network without considerable preparation time, plan to use the IMS Monitor and IMS line traces to investigate those IMS lines.

Detailed monitoring

VTAMPARS can be used to analyze the GTF SMS records and to tune the VTAM® buffer sizes. Sorting of DFSILTA0 and IMSPA Transit Log data by line or PTERM can isolate line-related problems for output queues to a single line or group of lines.

The IMS Monitor indicates line-related IWAITs, which require tuning of the IMS High Input/Output Pool (HIOP). Although HIOP is dynamically allocated, you might need to adjust either the buffer size definitions or the upper expansion limit. IMS Monitor output that is of interest for contention in communications are:

  • The Communication Summary report gives the mean elapsed time for transactions using a VTAM node. This time can be affected by activity in the message format buffer pool.

    The report also gives mean NON-IWAIT times. These can reflect device bottlenecks or problems in obtaining an input thread as described in Save area sets in DB/DC and DCCTL environments. The count of times that threads were not available is given in a separate report under the heading REPORTS. The line ‘Total Times ECBS waited for SAPS = nnn' gives the total. This figure is high if the monitor interval includes the time during which network initialization took place.

  • The Line Functions report shows, for each active node, the data transmission activity, as described in Collecting and interpreting IMS monitoring data. The TRANSMISSION BLKSIZE values show the number of characters passed by IMS to VTAM in OUTBUF buffers, a possible indicator of overloading of VTAM buffers. The TURNAROUND intervals can be used to identify inactive lines that might be eligible to distribute transaction entry.
  • The Communication IWAIT report gives details of IWAITs for each node.

To trace VTAM activity, you can use GTF (including the Buffer Trace) to examine the content of VTAM buffers as data enters and leaves VTAM buffers. You can also use the Storage Management trace to reflect the overall use VTAM buffers.

Ultimately, tuning of network problems probably requires one or more of the following actions:

  • Reduction of volumes of data transmitted. (See Choosing IMS options for performance.)
  • Adjustment of VTAM or NCP options and parameters.
  • Reconfiguration of the network to eliminate bottlenecks on multi-dropped lines.
  • Isolation and correction of hardware delays caused by device errors.