Neo: a learned query optimizer
© 2019, is held by the owner/author(s). 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....
| Main Authors: | , , , , , , , |
|---|---|
| Other Authors: | |
| Format: | Article |
| Language: | English |
| Published: |
VLDB Endowment
2021
|
| Online Access: | https://hdl.handle.net/1721.1/136597 |
| _version_ | 1826192540307554304 |
|---|---|
| author | Marcus, Ryan Negi, Parimarjan Mao, Hongzi Zhang, Chi Alizadeh, Mohammad Kraska, Tim Papaemmanouil, Olga Tatbul, Nesime |
| author2 | Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory |
| author_facet | Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory Marcus, Ryan Negi, Parimarjan Mao, Hongzi Zhang, Chi Alizadeh, Mohammad Kraska, Tim Papaemmanouil, Olga Tatbul, Nesime |
| author_sort | Marcus, Ryan |
| collection | MIT |
| description | © 2019, is held by the owner/author(s). 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. |
| first_indexed | 2024-09-23T09:19:56Z |
| format | Article |
| id | mit-1721.1/136597 |
| institution | Massachusetts Institute of Technology |
| language | English |
| last_indexed | 2024-09-23T09:19:56Z |
| publishDate | 2021 |
| publisher | VLDB Endowment |
| record_format | dspace |
| spelling | mit-1721.1/1365972023-09-06T20:17:17Z Neo: a learned query optimizer Marcus, Ryan Negi, Parimarjan Mao, Hongzi Zhang, Chi Alizadeh, Mohammad Kraska, Tim Papaemmanouil, Olga Tatbul, Nesime Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science © 2019, is held by the owner/author(s). 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. 2021-10-27T20:36:10Z 2021-10-27T20:36:10Z 2019 2020-11-23T19:06:08Z Article http://purl.org/eprint/type/ConferencePaper https://hdl.handle.net/1721.1/136597 en 10.14778/3342263.3342644 Proceedings of the VLDB Endowment Creative Commons Attribution-NonCommercial-NoDerivs License http://creativecommons.org/licenses/by-nc-nd/4.0/ application/pdf VLDB Endowment ACM |
| spellingShingle | Marcus, Ryan Negi, Parimarjan Mao, Hongzi Zhang, Chi Alizadeh, Mohammad Kraska, Tim Papaemmanouil, Olga Tatbul, Nesime Neo: a learned query optimizer |
| title | Neo: a learned query optimizer |
| title_full | Neo: a learned query optimizer |
| title_fullStr | Neo: a learned query optimizer |
| title_full_unstemmed | Neo: a learned query optimizer |
| title_short | Neo: a learned query optimizer |
| title_sort | neo a learned query optimizer |
| url | https://hdl.handle.net/1721.1/136597 |
| work_keys_str_mv | AT marcusryan neoalearnedqueryoptimizer AT negiparimarjan neoalearnedqueryoptimizer AT maohongzi neoalearnedqueryoptimizer AT zhangchi neoalearnedqueryoptimizer AT alizadehmohammad neoalearnedqueryoptimizer AT kraskatim neoalearnedqueryoptimizer AT papaemmanouilolga neoalearnedqueryoptimizer AT tatbulnesime neoalearnedqueryoptimizer |