Preference based Recommendation system deals with providing the service list to the users based on the preferences of the diverse users. Preferences play an important role in providing the appropriate services. User preferences are indicated by keywords and user-based collaborative algorithm is applied to generate the recommendations. In the current scenario, most of the recommendation systems suffer from issues such as scalability and efficiency in order to overcome these issues, the system is implemented on Hadoop. Hadoop is distributed computing platform that makes use of MapReduce framework. In this project, Hadoop is implemented on single node cluster. This system improves the scalability and accuracy over the existing systems.