SparkSQL+: Next-generation Query Planning over Spark
Binyang Dai, Qichen Wang, Ke Yi · 2023
We will demonstrate SparkSQL+, a SQL processing engine built on top of Spark. Unlike the vanilla SparkSQL that uses classical query plans, SparkSQL+ adopts some of the recently developed new query plans, including generalized hypertree decompositions(GHD), worst-case optimal join (WCOJ) algorithms, and conjunctive queries with comparisons (CQC). SparkSQL+ also provides a platform for users to explore different query plans for a given query through a web-based interface, and compare their performance with classical query plans on the same Spark core.