How often should the IDOC cache be run?
- Probably monthly
Would it be sensible to have a workflow that loops round each Production system to get the IDoc usage information?
- Only need one IDoc external data source since it contains the system name
The Cache IDoc Impact Data template reads SAP table EDIDC. We've narrowed down the columns but can we also filter by rows? The concern is runtime and size. It took days to read on six of our customers's production instances. Do we need to read the whole table to identify used partner profiles? Is there a recommended best practice or filter?
We have to do a full read on the table EDIDC to get the idocs that exist in there to tell which ones they are using; unfortunately we don't have a 'select distinct' option for the initial read of the table. We have to read it all then do the distinct using exec sql action. The only way I can see it speeding up is if the customer is happy to reduce the evaluation window by limiting the records using the create date/time fields or even the changed on date/time fields on the idocs in table EDIDC. The update date/time fields will change when the idoc has been interacted with (i.e. processed) so maybe this is a better way of identifying if that idoc type was 'used'. These fields are: CREDAT, CRETIM, UPDDAT and UPDTIM. You could add them into the Rows to Read criteria in the initial read sap table action for EDIDC.