Abstract. Broadcasting is a very effective technique to disseminate information to a massive number of clients when the data size is small. However, if the data size is large, the broadcast cycle may be long, and hence the access delay becomes a problem. Caching frequently accessed data at the client side can reduce the access latency and improve the bandwidth utilization. However, caching techniques may not perform well when the data are frequently updated. In this paper, we propose to apply different techniques (broadcasting and caching) to deal with different components of the data based on their update frequency. Compared to previous schemes, the proposed solution not only reduces the query latency, but also improves the throughput and the bandwidth utilization.