How "detect deleted" works in full aggregation?

[#1]

• This is for full aggregation with "Detect Deleted" checked in its task definition. For Delta aggregation, "Detect Deleted" will be ignored;

• The delete phase always happens after the primary aggregate phase during a aggregation;
• In the primary aggregate phase, no matter the optimization option is enabled or not, accounts that exist in source will always have their last refresh timestamp updated;

   

• When the aggregation starts, it gets the current timestamp ( Date startTimestamp = new Date() ) and when it gets to the delete phase, it does the comparison as below:

1. select *  
2. from spt_link  
3. where (last_refresh < startTimestamp  or last_refresh isnull) and
4. created < startTimestamp and
5. application = "source app"; 

Hope this helps with your understanding of how should-be-deleted accounts are detected.

For more details on aggregation workflow, please see Aggregation Data Flow.pdf

[kbeckwith]

How can we do this with delta.  The source always returns all records, but we check delta to eliminate redundant refreshes.   If I could also get deletes to work this way it would prevent creating another task for deletes. 

[shikida]

is there any way to intercept these records to be deleted and set a flag instead? If so, in which rule?

[prajanwal]

Hi shikida, were you able to find a solution to intercept deleting records?