Data loss in job sourcing from BW

I am running a batch job which pulls data from a SAP BW server. (Source- SAP bW ; Target - TeraData)

While i do a count of records in BW it comes out to be 1272984. Whereas if i run a full load of this job (in a isolated dataflow) i am able to load only 224408 records.

Can someone please explain me why would this happen? Do we need to commit data in Oracle or something.    

Note:- We are not using ABAP (R/3) flows to pull the data,