Recommender systems are valuable tools to provide service recommendations to the users. The data available online is growing rapidly because online activity of customers has grown rapidly. This has raised big data analysis problem for recommender systems as consumers of service demands better recommendations from the service providers. To process and analyze this large scale data the traditional service recommenders systems suffer the problem of scalability and inefficiency. Most service recommender systems lack a level of personalization which means the recommendations generated are not personalized for users. So, a user may not get the recommendations which he likes. In this paper we present a faster and better collaborative filtering recommender system technique then pre existing techniques. The problem of scalability is addressed by using Hadoop framework with OpenCL.