set JAVA_OPTS=-Xms500m -Xmx500m -XX:MaxNewSize=280m -XX:MaxPermSize=280m
could you not preprocess the dataset that contains your "array" field *before* trying to join it?
in this way you could actually run a 'select ID, col_ARRAY from urTable' load these into Talend denormalise it.
save into a temp table/file and then use it.
see if that helps.
so you want to create a 'col_array_label' using your 'col_array'
may i suggest:
get the distinct from your 'col_array'
select distinct col_array from urTable
then denormalise all of these
lookup their labels and create your 'col_array_label' output
inner join the output with your dataset
this should work pretty well if you have small universe (few combinations) in your col_array
*suggestion* - you may want to review this data structure and keep these 'arrays' in another table with a PK and refer to that instead of having it stored multiple times in your table; even on a reporting DB.