pemen wrote:

I need to have 5 source table in database with a lot of data in these tables.
It's not possible to us the standard DBInput for all tables because the lookup will be too big.
So the idea is to use different tELTInput connected to a tELTMap.
The tELTMap process the (big) query in the database and the output of this tELTMap go to
a standard tMap instead of tELTOutput. In this case the job can continue in the standard TALEND
process.

Is it possible ?

Hi Pierre-Eric,
ELT components generate SQL query to insert/update/delete rows on one or several  DB tables.
My proposal is to split your job on 3 sub job :
1/ The first subjob used 5 ELTInput connected to a tELTMap. The tELTMap connected  to a tELTOutput (the staging table).
2/ A subjob (called by the previous subjob thanks a "Then run" link) will read the stagging data (tDBInput) and will be connected to your tMap.
3/ The last subjob (called by the previous subjob thanks a "Then run" link) will delete the temporary table (truncate stagging table).