Managing logging for multiple Data Capture exit routines for a single EXIT= parameter

For example, if an EXIT= parameter is coded as follows:

EXIT=((PRODUCTA, KEY, DATA, CASCADE, LOG), (PRODUCTB, KEY, DATA, NONCASCADE, LOG))

when segments are inserted, updated, or deleted, IMS first records the change data per the exit routine specifications for PRODUCTA in one log record and then the change data for PRODUCTB’s specifications in another.

Separate change capture log records are required, because the products that might consume the change capture log records may not be able to tolerate some of the data logged for the exit options of another product. Use of multiple records with different exit names, as specified in the EXIT= parameter, allows products to only process log records written for their product.

But excessive logging can impact the performance of certain IMS processes. Depending on the products that consume the change capture log records, however, you may be able to take steps to reduce the total amount of log data being written by IMS. If your products can tolerate extra information in the log data, you can try one of the following options.

Option 1

If the products can tolerate logged data they do not need, and one product requests all the information the other requires, then the LOG attribute only needs to be included for the product that requests the superset of data needed. For example:

EXIT=((PRODUCTA,KEY, DATA, CASCADE, LOG), (PRODUCTB, KEY, DATA, NONCASCADE, NOLOG))

In the preceding example, both the PRODUCTA and PRODUCTB exit routines will be called with the information they request, but only one change capture log record will be written for an update. Note that if PRODUCTB reads that record, it will need to consume it, even though the exit name is PRODUCTA, and tolerate the log data for the CASCADE information.

Option 2

If the products can tolerate logged data they do not need, but neither requests a superset of information, consider specifying NOLOG on all the named exit routines and adding one with all required options and LOG but no exit name, as in this example:

EXIT=((PRODUCTA, CASCADE, NOLOG), (PRODUCTB, KEY, DATA, NONCASCADE, NOLOG),(*,KEY,DATA,CASCADE,LOG))

Here, the PRODUCTA and PRODUCTB exit routines will be called with the information they request, but there will be no change capture log records written using their exit names. Instead, there will be an unnamed record written that includes the superset of all data needed by both products. Note that in this example, if PRODUCTA reads that log record, it will need to ignore the KEY data, whereas PRODUCTB will need to ignore the CASCADE data.