We have a big batch campaign (with DB template) running every day. Every time we run the campaign the data in pxDecisionResults is growing, and it has reached more than 3 TB size till now. Customer does not want to store these interactions in this dataset for this specific campaign.
Is there any OOTB way ? or can we customize any OOTB rules to achieve this?
Hi, you can shorten the time to live for these records by using the dynamic system settings
ADMTimeoutForDelayedLearning for batch and it's companion
ADMShortTimeoutForDelayedLearning for real time containers.
There's a default grace period in Cassandra of 10 days prior to compaction tacking place and compaction will happen when a set amount of data is available, so, it may ( probably won’t ) not happen immediately.
Posted: 3 years ago
Posted: 30 Jan 2019 6:20 EST
Girish Kamath (kamag)
Software Architect, Decisioning Operations
In the dataflow that is executed for the campaign, on the strategy shape properties panel, you can set the execution mode to 'Make decision' instead of 'Make decision and store results for later response capture'. This way the strategy results will not be written to the pxDecisionResults dataset.
However, you should make an informed choice here. The pxDecisionResults dataset is vital for Pega's delayed learning capabilities to be leveraged. Not writing strategy results to it can mean you will not be able to leverage the delayed learning capabilities for your campaign.