I have a job with a tMatchGroup which matches customers together from various source systems by using name and adress, with levenshtein algorithm.
It works fine.
When a new customer is created in source system, i currently need to reload the job and the whole data are processed by the tMatchGroup.
If i have 10 000 customers, the job will process these 10 000 records together. But it is useless, I only need to match the new customer with the others.
Do i have to use another component ? Because It seems tMatchGroup use the whole table in input. (like a cartesian join)
Besides, I would like to keep the GID from the first launch, if i process the entire data, i will lost the GID, they will be replaced by new ones.
Or maybe I do not use the appropriate method...
Thanks for your help.