Tighter Upper Bounds for Join Cardinality Estimates

1 PROBLEM AND MOTIVATION Despite decades of research, modern database systems still struggle with multijoin queries. Users will often experience long wait times occurring with unpredictable frequency detracting from the usability of the system. In this work we develop a new method to tighten join cardinality upper bounds. The intention for these bounds is to assist the query optimizer (QO) in avoiding expensive physical join plans. Our approach is as follows: leveraging data sketching, and randomized hashing we generate and tighten theoretical join cardinality upper bounds. We outline our base data structures and methodology, and how these bounds may be introduced to a traditional QO framework as a new statistic for physical join plan selection. We evaluate the tightness of our bounds on GooglePlus community graphs and successfully generate degree of magnitude upper bounds even in the presence of multiway cyclic joins.