SkinnerDB: Regret-bounded Query Evaluation via Reinforcement Learning

Abstract

SkinnerDB uses reinforcement learning for reliable join ordering, exploiting an adaptive processing engine with specialized join algorithms and data structures. It maintains no data statistics and uses no cost or cardinality models. Also, it uses no training workloads nor does it try to link the current query to seemingly similar queries in the past. Instead, it uses reinforcement learning to learn optimal join orders from scratch during the execution of the current query. To that purpose, it divides the execution of a query into many small time slices. Different join orders are tried in different time slices. SkinnerDB merges result tuples generated according to different join orders until a complete query result is obtained. By measuring execution progress per time slice, it identifies promising join orders as execution proceeds. Along with SkinnerDB, we introduce a new quality criterion for query execution strategies. We upper-bound expected execution cost regret, i.e., the expected amount of execution cost wasted due to sub-optimal join order choices. SkinnerDB features multiple execution strategies that are optimized for that criterion. Some of them can be executed on top of existing database systems. For maximal performance, we introduce a customized execution engine, facilitating fast join order switching via specialized multi-way join algorithms and tuple representations. We experimentally compare SkinnerDB’s performance against various baselines, including MonetDB, Postgres, and adaptive processing methods. We consider various benchmarks, including the join order benchmark, TPC-H, and JCC-H, as well as benchmark variants with user-defined functions. Overall, the overheads of reliable join ordering are negligible compared to the performance impact of the occasional, catastrophic join order choice.

Online Sketch-based Query Optimization

“…Given the high variance among queries, these numbers have to be taken with a grain of salt since they may be dominated by a few complex queries with a large number of joins. Nonetheless, we follow prior art [5,54] and include them together with the aggregated workload statistics. As expected, COMPASS has the overall fastest runtime.…”

“…and SkinnerDB [54] re-run the query optimizer at runtime in the case of large differences between estimations and the true cardinalities. Wu et al [59] apply online sampling to correct the errors in the plans generated by the query optimizer.…”

Efficient and Effective Similar Subtrajectory Search with Deep Reinforcement Learning

“…For example, Zhang et al [42] and Li et al [17] use RL model for automatic DBMS tuning. Trummer et al [33] use RL to learn optimal join orders in the SkinnerDB system. Wang et al [36] design an effective RL-based algorithm for bipartite graph matching.…”

Neo: A Learned Query Optimizer

“…For example, recent work [18,57] showed how to exploit reinforcement learning for Eddies-style, fine-grained adaptive query processing. More recently, Trummer et al have proposed the SkinnerDB system, based on the idea of using regret bound as a quality measure while using reinforcement learning for dynamically improving the execution of an individual query in an adaptive query processing system [56]. Ortiz et al analyzed how state representations affect query optimization when using reinforcement learning [43].…”

“…Unlike join order selection, identifying join operators (e.g., hash join, merge join) and selecting indexes cannot be (entirely) reduced to cardinality estimation. Finally, Skin-nerDB showed that adaptive query processing strategies can benefit from reinforcement learning, but requires a specialized query execution engine, and cannot benefit from operator pipelining or other advanced parallelism models [56].…”