Efficient implementation techniques for the time index

A new indexing technique, time index, for improving the performance of certain classes of temporal queries was previously described by the author (see The 16th Conference on Very Large Databases (1990)). Three variations for implementing the time index efficiently are proposed and the performance of these three variations is compared with the performance of the original time index. Various parameters such as average lifetime of a version, average number of versions per object, block size and block clustering, and query time interval length are discussed. Simulation results show how these parameters affect the performance of the various implementation variations for the time index.<
>