Neo: A Learned Query Optimizer

By Ryan Marcus, Parimarjan Negi, Hongzi Mao, Chi Zhang, Mohammad Alizadeh, Tim Kraska, Olga Papaemmanouil, Nesime Tatbul. Proceedings of the VLDB Endowment, 2019

518 citations recorded by the source corpus.

Query optimization is one of the most challenging problems in database systems. Despite the progress made over the past decades, query optimizers remain extremely complex components that require a great deal of hand-tuning for specific workloads and datasets. Motivated by this shortcoming and inspired by recent advances in applying machine learning to data management challenges, we introduce Neo ( Neural Optimizer ), a novel learning-based query optimizer that relies on deep neural networks to generate query executions plans. Neo bootstraps its query optimization model from existing optimizers and continues to learn from incoming queries, building upon its successes and learning from its failures. Furthermore, Neo naturally adapts to underlying data patterns and is robust to estimation errors. Experimental results demonstrate that Neo, even when bootstrapped from a simple optimizer like PostgreSQL, can learn a model that offers similar performance to state-of-the-art commercial optimizers, and in some cases even surpass them.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Signal Processing

View Paper: https://www.semanticscholar.org/paper/5ba52bbe1101939c490a06cc0cf316a09000834e

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Signal Processing

Citations in the corpus

Learning scheduling algorithms for data processing clusters (2018, background): “1) has applications in problems over graphs, such as database query optimization [56] and hardware device placement [3].”
Bao: Making Learned Query Optimization Practical (2020, background): “changes, or risk becoming stale. Several proposed reinforcement learning techniques assume that both the data and the schema remain constant, and require complete retraining when this is not the case [29,39,41,47]. 3. Tail catastrophe. Recent work has shown that learning techniques can outperform tr”
ALEX: An Updatable Adaptive Learned Index (2019, methodology): “MLinotherDBcomponents:Machine learninghasbeen used to improve cardinality estimation [11, 17], query optimization [27], workload forecasting [25], multi-dimensional indexing [28], and data partitioning [36].”