Similar to the parallel join mechanism described in Chap. 23, aggregation operations can also be accelerated using parallelism and hash-based algorithms. In this chapter, we discuss how parallel aggregation is implemented in SanssouciDB. Note that multiple other ways to implement parallel aggregation are also conceivable. However, we focus on our parallel implementation using hashing and thread-local storage.
- [Pla11]H. Plattner, D.B. Sanssouci, An in-memory database for processing enterprise workloads, in BTW, LNI, ed. by T. Härder, W. Lehner, B. Mitschang, H. Schöning, H. Schwarz, vol. 180 (GI, 2011), pp. 2–21Google Scholar