Interpreting temporary storage statistics
If a data item is written to temporary storage (using WRITEQ TS), a temporary storage queue is built and temporary storage statistics are produced.
The following statistics might require the actions described:
- Writes more than control interval
- The number of writes of records whose length was greater than the control interval (CI) size of the TS data set. Use this value to adjust the CI size. If the reported value is large, increase the CI size. If the value is zero, consider reducing the CI size until a small value is reported.
- Times aux. storage exhausted
- The number of situations where one or more transactions might have been suspended because of a NOSPACE condition, or might have been forced to abend (by using a HANDLE CONDITION NOSPACE command, or using RESP on the WRITEQ TS command, or WRITEQ TS NOSUSPEND command). If this item appears in the statistics, increase the size of the temporary storage data set.
- Buffer writes
- The number of WRITEs to the temporary storage data set. This includes both WRITEs required for recovery and WRITEs required when the buffer is needed to accommodate another control interval (CI). To minimize input/output activity caused by the second situation, increase buffer allocation. Use the system initialization parameter, TS=(b,s), where b is the number of buffers and s is the number of strings.
- Peak number of strings in use
- The peak number of concurrent I/O operations to the data set. If this is significantly less than the number of strings specified in the TS system initialization parameter, consider reducing the SIT value to approach this number.
- Times string wait occurred
- The number of input/output requests that were queued because no strings were available. If this value is not zero, consider increasing the number of strings. For details about adjusting the size of the TS data set and the number of strings and buffers, see Storage calculations for temporary storage data sharing.