Recommender systems, now used in many websites, have a critical role in e-commerce systems. Among many techniques for recommender systems so far, collaborative filtering has proved to be one of the best solutions. However, there still remain some problems for collaborative filtering recommender systems, i.e. the scalability problem, which is the most critical one. In this paper, we focus on the scalability problem and present some algorithmic elements to improve the scalability of recommender systems. Then we show the trade-offs which the newly introduced elements bring about.