Title: Temporal Databases and Parallelism ================================== Speaker: Thomas Zurek Date: Wednesday 31 May 1995 Time: 12:30 Location JCMB 3316 ------------------------------------------------------------------------ Abstract ======== This talk will give a short introduction to parallel and temporal databases led by simple examples. The combination of these two areas seems to be quite promising: parallelism could solve some of the performance gaps imposed by temporal query processing. After introducing temporal databases and the types of parallelism that can be exploited, an example will be used to show how a traditional join - one of the most expensive database operations - can be processed by partitioning the data over several processors. What works nicely for the traditional join, however, gets more difficult in the case of the temporal join: data has to be replicated and produces a significant overhead. By changing the original example it can be shown that the balance has to be found between minimizing the overhead and maximizing parallelism (the so called "min-max-dilemma"). I will conclude the talk with presenting several possibilities for tackling this problem and hope to get some feedback about what you consider as the most promising way.