Distributed data is universal in present day in- formation driven applications. As there are n number of sources for data today, the natural difficulty is to identify how to combine the data more effectively upon Organizational boundaries while using the data to maximum. As utilizing onlythe local data gives suboptimal use, different techniques must be developed for privacy preserving collaborating knowledge discovery. For the large scale data sets the existing system i.e., cryptography based function for privacy preserving data mining is very slow and effective to face present day big data challenge. Earlier work on random decision trees shows that it is possible to develop more accurate and effective with less cost. As studied we can conclude that RDTs can naturally accept a parallel and fully distributed architecture and accordingly generate protocols to implement privacy preserving RDTs that ensure basic and efficient distributed privacy preserving knowledgediscovery.