Bao

Recent efforts applying machine learning techniques to query optimization have shown few practical gains due to substantive training overhead, inability to adapt to changes, and poor tail performance. Motivated by these difficulties, we introduce Bao (the Bandit optimizer). Bao takes advantage of the wisdom built into existing query optimizers by providing per-query optimization hints. Bao combines modern tree convolutional neural networks with Thompson sampling, a well-studied reinforcement learning algorithm. As a result, Bao automatically learns from its mistakes and adapts to changes in query workloads, data, and schema. Experimentally, we demonstrate that Bao can quickly learn strategies that improve end-to-end query execution performance, including tail latency, for several workloads containing longrunning queries. In cloud environments, we show that Bao can offer both reduced costs and better performance compared with a commercial system.

• Publication year

  2022

• Venue

  SIGMOD record

• Publication date

  2022-05-31

• Fields of study

  Computer Science

• Identifiers
  DOI 10.1145/3542700.3542703
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Making Learned Query Optimization Practical

  2022influential reference
• Steering Query Optimizers: A Practical Take on Big Data Workloads

  2021cited by this paper
• Bao: Making Learned Query Optimization Practical

  2020cited by this paper
• Cost-Guided Cardinality Estimation: Focus Where it Matters

  2020cited by this paper
• Research challenges in deep reinforcement learning-based join query optimization

  2020cited by this paper
• An End-to-End Learning-based Cost Estimator

  2019cited by this paper
• AI Meets AI: Leveraging Query Executions to Improve Index Recommendations

  2019cited by this paper
• An Empirical Analysis of Deep Learning for Cardinality Estimation

  2019cited by this paper
• Neo: A Learned Query Optimizer

  2019influential reference
• Termite: a system for tunneling through heterogeneous data

  2019cited by this paper
• Deep Unsupervised Cardinality Estimation

  2019cited by this paper
• Learning to Optimize Join Queries With Deep Reinforcement Learning

  2018cited by this paper
• Cuttlefish: A Lightweight Primitive for Adaptive Query Processing

  2018cited by this paper
• Deep Reinforcement Learning for Join Order Enumeration

  2018cited by this paper
• Learning State Representations for Query Optimization with Deep Reinforcement Learning

  2018cited by this paper
• Query2Vec: An Evaluation of NLP Techniques for Generalized Workload Analytics

  2018cited by this paper
• The three pillars of machine programming

  2018cited by this paper
• Deep Contextual Multi-armed Bandits

  2018cited by this paper
• SkinnerDB: Regret-Bounded Query Evaluation via Reinforcement Learning

  2018cited by this paper
• Database-Agnostic Workload Management

  2018cited by this paper
• Learned Cardinalities: Estimating Correlated Joins with Deep Learning

  2018cited by this paper
• QuickSel: Quick Selectivity Learning with Mixture Models

  2018cited by this paper
• 基於Google Cloud Platform設計高效能日誌分析平台之研究

  2017cited by this paper
• The Case for Learned Index Structures

  2017cited by this paper
• Cardinality estimation using neural networks

  2015cited by this paper
• A Survey on Contextual Multi-armed Bandits

  2015cited by this paper
• How Good Are Query Optimizers, Really?

  2015cited by this paper
• Convolutional Neural Networks over Tree Structures for Programming Language Processing

  2014influential reference
• Contender: A Resource Modeling Approach for Concurrent Query Performance Prediction

  2014cited by this paper
• On Predictive Modeling for Optimizing Transaction Execution in Parallel OLTP Systems

  2011cited by this paper
• An Empirical Evaluation of Thompson Sampling

  2011influential reference
• A Reinforcement Learning Approach for Adaptive Query Processing

  2008cited by this paper
• The Need for Biases in Learning Generalizations

  2007cited by this paper
• DeWitt Clauses: Can We Protect Purchasers without Hurting Microsoft?

  2006cited by this paper
• LEO - DB2's LEarning Optimizer

  2001cited by this paper
• Query optimization

  1996cited by this paper
• Is query optimization a 'solved' problem?

  1989cited by this paper
• Access path selection in a relational database management system

  1979cited by this paper
• ON THE LIKELIHOOD THAT ONE UNKNOWN PROBABILITY EXCEEDS ANOTHER IN VIEW OF THE EVIDENCE OF TWO SAMPLES

  1933cited by this paper
• 2015 Ieee International Conference on Big Data (big Data) Learning to Accurately Count with Query-driven Predictive Analytics

  year unknowncited by this paper

• No citing papers are available for this paper.