Logically clustered architectures for networked databases

By effectively harnessing networked computing resources, the two-tier client-server model has been used to support shared data access. In systems based on this approach, the database servers often become performance bottlenecks when the number of concurrent users is large. Client data caching techniques have been proposed in order to ease resource contention at the servers. The key theme of these techniques is the exploitation of user data access locality. In this paper, we propose a three-tiered model that takes advantage of such data access locality to furnish a much more scalable system. Groups of clients that demonstrate similarities in their data access behavior are logically clustered together. Each such group of clients is handled by an Intermediate Cluster Manager (ICM) that acts as a cluster-wide directory service and cache manager. Clients within the same cluster are now capable of sharing data among themselves without interacting with the server(s). This results in reduced server load and allows the support of a much larger number of clients. Through prototyping and experimentation, we show that the logical clustering of clients, and the introduction of the ICM layer, significantly improve system scalability as well as transaction response times. Logical clusters, consisting of clients with similar data access patterns, are identified with the help of both a greedy algorithm and a genetic algorithm. For the latter, we have developed an encoding scheme and its corresponding operators.