Extrapartition transient data considerations
Extrapartition destinations are, in practice, sequential data sets where CICS® uses the QSAM PUT LOCATE or PUT MOVE commands.
The main performance factor to note is the possibility of operating
system waits; that is, the complete CICS region
waits for the I/O completion. A lengthy wait can occur for one of
the following reasons:
- No buffer space is available.
- Secondary space is allocation.
- Volume (extent) switching is available.
- The data set has been opened or closed dynamically.
- A forced end of the volume has been caused by the application.
- The data set is defined on a physical printer and the printer has run out of paper.
- A RESERVE command has been issued for another data set on the same volume.
Therefore, try to eliminate or minimize the occurrences of CICS region
waits by:
- Having sufficient buffering and blocking of the output data set
- Avoiding volume switching by initially allocating sufficient space
- Avoiding dynamic OPEN or CLOSE actions during peak periods.
An alternative method of implementing sequential data sets is to employ a CICS user journal. Table 1 summarizes the differences between these two methods.
| Extrapartition TD | User Journal |
|---|---|
| Region (CICS) may wait | Task waits |
| Buffer location: In MVS™ storage | Buffer location: In DSA |
| Number of buffers: 1 - 32767 | 2 buffers |
| Input or output | Both input and output, but tasks may wait |
| Accessible by multiple tasks |
|
The approximate calculations for performance costs in
extrapartition TD queues do not include any I/O cost. An I/O operation
for a physically sequential file costs approximately 7 K and occurs
in the following situations:
- When attempting to write an item that does not fit in any buffer.
- When reading an item that is not in the buffer.
- When reading data from DASD and there is no available buffer space. If this situation occurs, the least recently used buffer must first be written out.
Extrapartition TD queues are nonrecoverable.
| WRITEQ | READQ |
|---|---|
| 1.2 | 1.0 |