Publicação

Query optimizers based on machine learning techniques

Ver documento

Detalhes bibliográficos
Resumo:Query optimizers are considered one of the most relevant and sophisticated components in a database management system. However, despite currently producing nearly optimal results, optimizers rely on statistical estimates and heuristics to reduce the search space of alternative execution plans for a single query. As a result, for more complex queries, errors may grow exponentially, often translating into sub-optimal plans resulting in less than ideal performance. Recent advances in machine learning techniques have opened new opportunities for many of the existing problems related to system optimization. This document proposes a solution built on top of PostgreSQL that learns to select the most efficient set of optimizer strategy settings for a particular query. Instead of depending entirely on the optimizer’s estimates to compare different plans under different configurations, it relies on a greedy selection algorithm that supports several types of predictive modeling techniques, from more traditional modeling techniques to a deep learning approach. The system is evaluated experimentally with the standard TPC-H and Join Order ing Benchmark workloads to measure the cost and benefits of adding machine learning capabilities to traditional query optimizers.
Autores principais:Souto, Rui Pedro Sousa Rodrigues do
Assunto:Database tuning Machine learning Query optimization Aprendizagem automática Otimização de queries Tuning de base de dados
Ano:2021
País:Portugal
Tipo de documento:dissertação de mestrado
Tipo de acesso:acesso aberto
Instituição associada:Universidade do Minho
Idioma:inglês
Origem:RepositóriUM - Universidade do Minho
_version_ 1866269718653960192
author Souto, Rui Pedro Sousa Rodrigues do
author_facet Souto, Rui Pedro Sousa Rodrigues do
author_role author
contributor_name_str_mv Coelho, Fábio André Castanheira Luís
Vilaça, Ricardo Manuel Pereira
Universidade do Minho
country_str PT
creators_json_txt [{\"Person.name\":\"Souto, Rui Pedro Sousa Rodrigues do\"}]
datacite.contributors.contributor.contributorName.fl_str_mv Coelho, Fábio André Castanheira Luís
Vilaça, Ricardo Manuel Pereira
Universidade do Minho
datacite.creators.creator.creatorName.fl_str_mv Souto, Rui Pedro Sousa Rodrigues do
datacite.date.Accepted.fl_str_mv 2021-10-27T00:00:00Z
datacite.date.available.fl_str_mv 2023-04-27T15:49:20Z
datacite.date.embargoed.fl_str_mv 2023-04-27T15:49:20Z
datacite.rights.fl_str_mv http://purl.org/coar/access_right/c_abf2
datacite.subjects.subject.fl_str_mv Database tuning
Machine learning
Query optimization
Aprendizagem automática
Otimização de queries
Tuning de base de dados
datacite.titles.title.fl_str_mv Query optimizers based on machine learning techniques
dc.contributor.none.fl_str_mv Coelho, Fábio André Castanheira Luís
Vilaça, Ricardo Manuel Pereira
Universidade do Minho
dc.creator.none.fl_str_mv Souto, Rui Pedro Sousa Rodrigues do
dc.date.Accepted.fl_str_mv 2021-10-27T00:00:00Z
dc.date.available.fl_str_mv 2023-04-27T15:49:20Z
dc.date.embargoed.fl_str_mv 2023-04-27T15:49:20Z
dc.format.none.fl_str_mv application/pdf
dc.identifier.none.fl_str_mv https://hdl.handle.net/1822/84186
dc.language.none.fl_str_mv eng
dc.rights.cclincense.fl_str_mv http://creativecommons.org/licenses/by/4.0/
dc.rights.none.fl_str_mv http://purl.org/coar/access_right/c_abf2
dc.rights.rights.copyright.fl_str_mv openAccess
dc.subject.none.fl_str_mv Database tuning
Machine learning
Query optimization
Aprendizagem automática
Otimização de queries
Tuning de base de dados
dc.title.fl_str_mv Query optimizers based on machine learning techniques
dc.type.none.fl_str_mv http://purl.org/coar/resource_type/c_bdcc
description Query optimizers are considered one of the most relevant and sophisticated components in a database management system. However, despite currently producing nearly optimal results, optimizers rely on statistical estimates and heuristics to reduce the search space of alternative execution plans for a single query. As a result, for more complex queries, errors may grow exponentially, often translating into sub-optimal plans resulting in less than ideal performance. Recent advances in machine learning techniques have opened new opportunities for many of the existing problems related to system optimization. This document proposes a solution built on top of PostgreSQL that learns to select the most efficient set of optimizer strategy settings for a particular query. Instead of depending entirely on the optimizer’s estimates to compare different plans under different configurations, it relies on a greedy selection algorithm that supports several types of predictive modeling techniques, from more traditional modeling techniques to a deep learning approach. The system is evaluated experimentally with the standard TPC-H and Join Order ing Benchmark workloads to measure the cost and benefits of adding machine learning capabilities to traditional query optimizers.
dirty 0
eu_rights_str_mv openAccess
format masterThesis
fulltext.url.fl_str_mv https://prod-dspace.uminho.pt/bitstreams/bc35ee58-2ed1-41cb-b0e9-b192d55c69b5/download
id rum_e88e73efc819a6bf7d3723ff385a88e1
identifier.url.fl_str_mv https://hdl.handle.net/1822/84186
instacron_str repositorium
institution Universidade do Minho
instname_str Universidade do Minho
language eng
network_acronym_str rum
network_name_str RepositóriUM - Universidade do Minho
oai_identifier_str oai:repositorium.uminho.pt:1822/84186
organization_str_mv urn:organizationAcronym:repositorium
person_str_mv Souto, Rui Pedro Sousa Rodrigues do
publishDate 2021
reponame_str RepositóriUM - Universidade do Minho
repository_id_str urn:repositoryAcronym:rum
service_str_mv urn:repositoryAcronym:rum
spelling engporQuery optimizers are considered one of the most relevant and sophisticated components in a database management system. However, despite currently producing nearly optimal results, optimizers rely on statistical estimates and heuristics to reduce the search space of alternative execution plans for a single query. As a result, for more complex queries, errors may grow exponentially, often translating into sub-optimal plans resulting in less than ideal performance. Recent advances in machine learning techniques have opened new opportunities for many of the existing problems related to system optimization. This document proposes a solution built on top of PostgreSQL that learns to select the most efficient set of optimizer strategy settings for a particular query. Instead of depending entirely on the optimizer’s estimates to compare different plans under different configurations, it relies on a greedy selection algorithm that supports several types of predictive modeling techniques, from more traditional modeling techniques to a deep learning approach. The system is evaluated experimentally with the standard TPC-H and Join Order ing Benchmark workloads to measure the cost and benefits of adding machine learning capabilities to traditional query optimizers.application/pdfporQuery optimizers based on machine learning techniquesSouto, Rui Pedro Sousa Rodrigues doCoelho, Fábio André Castanheira LuísVilaça, Ricardo Manuel PereiraHostingInstitutionOrganizationalUniversidade do Minhoe-mailmailto:repositorium@usdb.uminho.ptrepositorium@usdb.uminho.ptURNurn:tid:2032593352023-04-27T15:49:20Z2021-10-272021-072021-10-27T00:00:00ZHandlehttps://hdl.handle.net/1822/84186http://purl.org/coar/access_right/c_abf2open accessDatabase tuningMachine learningQuery optimizationAprendizagem automáticaOtimização de queriesTuning de base de dados1674965 bytesliteraturehttp://purl.org/coar/resource_type/c_bdccmaster thesis2021-10-27http://creativecommons.org/licenses/by/4.0/openAccesshttp://purl.org/coar/access_right/c_abf2application/pdffulltexthttps://prod-dspace.uminho.pt/bitstreams/bc35ee58-2ed1-41cb-b0e9-b192d55c69b5/download
spellingShingle Query optimizers based on machine learning techniques
Souto, Rui Pedro Sousa Rodrigues do
Database tuning
Machine learning
Query optimization
Aprendizagem automática
Otimização de queries
Tuning de base de dados
status SINGLETON
subject.fl_str_mv Database tuning
Machine learning
Query optimization
Aprendizagem automática
Otimização de queries
Tuning de base de dados
title Query optimizers based on machine learning techniques
title_full Query optimizers based on machine learning techniques
title_fullStr Query optimizers based on machine learning techniques
title_full_unstemmed Query optimizers based on machine learning techniques
title_short Query optimizers based on machine learning techniques
title_sort Query optimizers based on machine learning techniques
topic Database tuning
Machine learning
Query optimization
Aprendizagem automática
Otimização de queries
Tuning de base de dados
topic_facet Database tuning
Machine learning
Query optimization
Aprendizagem automática
Otimização de queries
Tuning de base de dados
url https://hdl.handle.net/1822/84186
visible 1