suresh gorantla

• Dec 29th, 2005

 

Source based commit will commit the data into target based on commit interval.so,for every 10,000 rows it will commit into target.

Target based commit will commit the data into target based on buffer size of the target.i.e., it commits the data into target when ever the buffer fills.Let us assume that the buffer size is 6,000.So,for every 6,000 rows it commits the data.

vivek1708

 Profile Answers by vivek1708 

• May 11th, 2008

 

I dont think above written statement is correct

Myunderstanding is that

for source based commit: a session as soon as it reads 10k rows from the source it performs a commit
and similarly for tbc: as soon as it write 10k rows to targt it does a commit

prade

 Profile Answers by prade Questions by prade

• Jun 26th, 2008

 

In case of a source based commit, the integration services generates a commit row from the active sources and writes the rows into the target. The number of rows processed from the "active source" is the driving factor in commit intervals. If the "active source" processes 10,000 rows out of which 2000 are discarded due to mapping logic then only 8000 rows are committed else all 10,000 are committed. If the session completes successfully then the commit points are 10k, 20k, 30k, 40k and 50k.

It should be noted here that "active sources" are not always the source qualifiers and it is the transformation to the target with all input transaformation scope in the line of the transactional unit.


Incase of a mapping with two transactional units and a single SQ
SQ----> AG ------> Target1
    ----->FILTER----> target2

for target1 AG is the commit source whereas for target2, Filter is the commit source