Client-Server Caching Revisited

The caching of data and/or locks at client workstations is an eeective technique for improving the performance of a client-server database system. This paper extends an earlier performance study of client-server caching in several ways. The rst is a re-examination of heuristics for deciding dynamically between propagating changes or invalidating remote copies of data pages in order to maintain cache consistency. The second is a study of the \Callback Locking" family of caching algorithms. These algorithms are of interest because they provide an alternative to the optimistic techniques used in the earlier study and because they have recently begun to nd use in commercial systems. In addition, the performance of the caching algorithms is examined in light of current trends in processor and network speeds and in the presence of data contention.