While configuring CDH context dictionary we can use DB as a data store or a combination of DDS and DB as data store.
Now considering the information given in below link , it seemed that using a DDS will give better performance, hence we went for the option of using a combination of DDS and DB in the context dictionary.
However later doing a performance test we found that doing compose of two DDS in a data flow is 10 times slower than doing a compose of two DB tables in a Data flow. Now this finding goes against the info given in the above link.
We will like some suggestion on what should be the optimum choice of data store.
Kindly revert ASAP as our current implementation timeline is heavily dependent on it.