Rohit - there really are not any demo apps built against the SR3 release of Qlik, in that version the only DD feature that was implemented was the ability to pre-aggregate fields, using the MEASURE keyword, in the src system then load the result into memory. This was a great start for some simple use cases, but SR5 is where the majority of the key functionality was delivered.
There are 3 basic use cases Direct Discovery is designed to exploit, and the general benefit with each is that you can provide access to more data than can comfortable fit into memory, because while Qlikview scales very well it can't keep pace with the growth in data size, from millions of rows, to billions and beyond. The use cases are:
1) Drill to Detail - For a large table with say 100+ columns, load only the critical fields into memory, and leave the majority of the fields in the source system. This saves as much as 90% in terms of memory space, but still allows the user to explore all of the data
2) Historical comparison - Provide access to deep historical data, again without loading everything into memory. The QLik App can focus on analysis of the last 6mos of data, but the user can also compare to 2yrs or 10yrs ago.
3) Realtime Analysis - Because Direct Discovery can reach out to source systems with real-time, adhoc queries, Qlikview can now combine the best of in-memory performance with direct query, answering the question "what is happening now?"
