{"id":"https://openalex.org/W4372262787","doi":"https://doi.org/10.1145/3552326.3587445","title":"Lyra: Elastic Scheduling for Deep Learning Clusters","display_name":"Lyra: Elastic Scheduling for Deep Learning Clusters","publication_year":2023,"publication_date":"2023-05-05","ids":{"openalex":"https://openalex.org/W4372262787","doi":"https://doi.org/10.1145/3552326.3587445"},"language":"en","primary_location":{"id":"doi:10.1145/3552326.3587445","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3552326.3587445","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3552326.3587445","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Eighteenth European Conference on Computer Systems","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://dl.acm.org/doi/pdf/10.1145/3552326.3587445","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5100730410","display_name":"Jiamin Li","orcid":"https://orcid.org/0000-0001-8110-2436"},"institutions":[{"id":"https://openalex.org/I168719708","display_name":"City University of Hong Kong","ror":"https://ror.org/03q8dnn23","country_code":"HK","type":"education","lineage":["https://openalex.org/I168719708"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Jiamin Li","raw_affiliation_strings":["City University of Hong Kong, Hong Kong, Hong Kong"],"raw_orcid":"https://orcid.org/0000-0001-8110-2436","affiliations":[{"raw_affiliation_string":"City University of Hong Kong, Hong Kong, Hong Kong","institution_ids":["https://openalex.org/I168719708"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5022420830","display_name":"Hong Xu","orcid":"https://orcid.org/0000-0002-9359-9571"},"institutions":[{"id":"https://openalex.org/I177725633","display_name":"Chinese University of Hong Kong","ror":"https://ror.org/00t33hh48","country_code":"HK","type":"education","lineage":["https://openalex.org/I177725633"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Hong Xu","raw_affiliation_strings":["The Chinese University of Hong Kong, Hong Kong, Hong Kong"],"raw_orcid":"https://orcid.org/0000-0002-9359-9571","affiliations":[{"raw_affiliation_string":"The Chinese University of Hong Kong, Hong Kong, Hong Kong","institution_ids":["https://openalex.org/I177725633"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5028120333","display_name":"Yibo Zhu","orcid":"https://orcid.org/0000-0002-9113-2660"},"institutions":[{"id":"https://openalex.org/I1291425158","display_name":"Google (United States)","ror":"https://ror.org/00njsd438","country_code":"US","type":"company","lineage":["https://openalex.org/I1291425158","https://openalex.org/I4210128969"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Yibo Zhu","raw_affiliation_strings":["Google, Kirkland, United States of America"],"raw_orcid":"https://orcid.org/0000-0002-9113-2660","affiliations":[{"raw_affiliation_string":"Google, Kirkland, United States of America","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5090569171","display_name":"Zherui Liu","orcid":"https://orcid.org/0009-0002-4910-6095"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zherui Liu","raw_affiliation_strings":["ByteDance Inc., Beijing, China"],"raw_orcid":"https://orcid.org/0009-0002-4910-6095","affiliations":[{"raw_affiliation_string":"ByteDance Inc., Beijing, China","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5054205326","display_name":"Chuanxiong Guo","orcid":"https://orcid.org/0000-0002-0730-8468"},"institutions":[{"id":"https://openalex.org/I174135032","display_name":"Bellevue College","ror":"https://ror.org/05gr4yv49","country_code":"US","type":"education","lineage":["https://openalex.org/I174135032"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Chuanxiong Guo","raw_affiliation_strings":["Non affiliated, Bellevue, United States of America"],"raw_orcid":"https://orcid.org/0000-0002-0730-8468","affiliations":[{"raw_affiliation_string":"Non affiliated, Bellevue, United States of America","institution_ids":["https://openalex.org/I174135032"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5100390514","display_name":"Cong Wang","orcid":"https://orcid.org/0000-0003-0547-315X"},"institutions":[{"id":"https://openalex.org/I168719708","display_name":"City University of Hong Kong","ror":"https://ror.org/03q8dnn23","country_code":"HK","type":"education","lineage":["https://openalex.org/I168719708"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Cong Wang","raw_affiliation_strings":["City University of Hong Kong, Hong Kong, Hong Kong"],"raw_orcid":"https://orcid.org/0000-0003-0547-315X","affiliations":[{"raw_affiliation_string":"City University of Hong Kong, Hong Kong, Hong Kong","institution_ids":["https://openalex.org/I168719708"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":4,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":16.4844,"has_fulltext":true,"cited_by_count":51,"citation_normalized_percentile":{"value":0.99392355,"is_in_top_1_percent":true,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":90,"max":100},"biblio":{"volume":null,"issue":null,"first_page":"835","last_page":"850"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10101","display_name":"Cloud Computing and Resource Management","score":0.9998000264167786,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10101","display_name":"Cloud Computing and Resource Management","score":0.9998000264167786,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10273","display_name":"IoT and Edge/Fog Computing","score":0.9987999796867371,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10715","display_name":"Distributed and Parallel Computing Systems","score":0.9973000288009644,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/server","display_name":"Server","score":0.7704732418060303},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7503484487533569},{"id":"https://openalex.org/keywords/scheduling","display_name":"Scheduling (production processes)","score":0.5475162863731384},{"id":"https://openalex.org/keywords/heuristics","display_name":"Heuristics","score":0.5465240478515625},{"id":"https://openalex.org/keywords/job-scheduler","display_name":"Job scheduler","score":0.49941349029541016},{"id":"https://openalex.org/keywords/cluster","display_name":"Cluster (spacecraft)","score":0.4945325553417206},{"id":"https://openalex.org/keywords/distributed-computing","display_name":"Distributed computing","score":0.4911775290966034},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.47641003131866455},{"id":"https://openalex.org/keywords/job-queue","display_name":"Job queue","score":0.43737873435020447},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.43323737382888794},{"id":"https://openalex.org/keywords/queue","display_name":"Queue","score":0.4329250156879425},{"id":"https://openalex.org/keywords/queueing-theory","display_name":"Queueing theory","score":0.4247759282588959},{"id":"https://openalex.org/keywords/computer-cluster","display_name":"Computer cluster","score":0.41256529092788696},{"id":"https://openalex.org/keywords/computer-network","display_name":"Computer network","score":0.28809261322021484},{"id":"https://openalex.org/keywords/operating-system","display_name":"Operating system","score":0.28593844175338745},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.2827749252319336},{"id":"https://openalex.org/keywords/mathematical-optimization","display_name":"Mathematical optimization","score":0.2547959089279175},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.09573942422866821}],"concepts":[{"id":"https://openalex.org/C93996380","wikidata":"https://www.wikidata.org/wiki/Q44127","display_name":"Server","level":2,"score":0.7704732418060303},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7503484487533569},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.5475162863731384},{"id":"https://openalex.org/C127705205","wikidata":"https://www.wikidata.org/wiki/Q5748245","display_name":"Heuristics","level":2,"score":0.5465240478515625},{"id":"https://openalex.org/C111873713","wikidata":"https://www.wikidata.org/wiki/Q1641413","display_name":"Job scheduler","level":3,"score":0.49941349029541016},{"id":"https://openalex.org/C164866538","wikidata":"https://www.wikidata.org/wiki/Q367351","display_name":"Cluster (spacecraft)","level":2,"score":0.4945325553417206},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.4911775290966034},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.47641003131866455},{"id":"https://openalex.org/C171627638","wikidata":"https://www.wikidata.org/wiki/Q6206744","display_name":"Job queue","level":4,"score":0.43737873435020447},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.43323737382888794},{"id":"https://openalex.org/C160403385","wikidata":"https://www.wikidata.org/wiki/Q220543","display_name":"Queue","level":2,"score":0.4329250156879425},{"id":"https://openalex.org/C22684755","wikidata":"https://www.wikidata.org/wiki/Q847526","display_name":"Queueing theory","level":2,"score":0.4247759282588959},{"id":"https://openalex.org/C29140674","wikidata":"https://www.wikidata.org/wiki/Q206637","display_name":"Computer cluster","level":2,"score":0.41256529092788696},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.28809261322021484},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.28593844175338745},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.2827749252319336},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.2547959089279175},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.09573942422866821}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3552326.3587445","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3552326.3587445","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3552326.3587445","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Eighteenth European Conference on Computer Systems","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.1145/3552326.3587445","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3552326.3587445","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3552326.3587445","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Eighteenth European Conference on Computer Systems","raw_type":"proceedings-article"},"sustainable_development_goals":[{"score":0.5,"id":"https://metadata.un.org/sdg/8","display_name":"Decent work and economic growth"}],"awards":[{"id":"https://openalex.org/G1495595508","display_name":null,"funder_award_id":"5501329","funder_id":"https://openalex.org/F4320322942","funder_display_name":"Chinese University of Hong Kong"},{"id":"https://openalex.org/G5248485515","display_name":null,"funder_award_id":"5501517","funder_id":"https://openalex.org/F4320322942","funder_display_name":"Chinese University of Hong Kong"},{"id":"https://openalex.org/G6552712453","display_name":null,"funder_award_id":"8601677","funder_id":"https://openalex.org/F4320322942","funder_display_name":"Chinese University of Hong Kong"},{"id":"https://openalex.org/G7255282877","display_name":null,"funder_award_id":"4937007","funder_id":"https://openalex.org/F4320322942","funder_display_name":"Chinese University of Hong Kong"},{"id":"https://openalex.org/G8059028106","display_name":null,"funder_award_id":"4937008","funder_id":"https://openalex.org/F4320322942","funder_display_name":"Chinese University of Hong Kong"}],"funders":[{"id":"https://openalex.org/F4320322942","display_name":"Chinese University of Hong Kong","ror":"https://ror.org/00t33hh48"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4372262787.pdf","grobid_xml":"https://content.openalex.org/works/W4372262787.grobid-xml"},"referenced_works_count":22,"referenced_works":["https://openalex.org/W90213227","https://openalex.org/W2018162709","https://openalex.org/W2060393849","https://openalex.org/W2079916778","https://openalex.org/W2105947650","https://openalex.org/W2112319566","https://openalex.org/W2141992894","https://openalex.org/W2156077332","https://openalex.org/W2194775991","https://openalex.org/W2289698129","https://openalex.org/W2335814492","https://openalex.org/W2535374105","https://openalex.org/W2724904824","https://openalex.org/W2798515322","https://openalex.org/W2899071864","https://openalex.org/W2919115771","https://openalex.org/W2969388332","https://openalex.org/W2975712713","https://openalex.org/W2982157693","https://openalex.org/W3022298203","https://openalex.org/W3106803824","https://openalex.org/W4288079579"],"related_works":["https://openalex.org/W2546696010","https://openalex.org/W1992741870","https://openalex.org/W2921990381","https://openalex.org/W2366792704","https://openalex.org/W2972076670","https://openalex.org/W1692590824","https://openalex.org/W42309186","https://openalex.org/W4301973339","https://openalex.org/W1970954203","https://openalex.org/W4221063637"],"abstract_inverted_index":{"Organizations":[0],"often":[1,37],"build":[2],"separate":[3,13],"training":[4,35,68,78],"and":[5,11,89,125,188,204,211,213],"inference":[6,24,65],"clusters":[7,25],"for":[8,22,67,164],"deep":[9],"learning,":[10],"use":[12],"schedulers":[14],"to":[15,20,42,54,62,82,95,103,108,119,122,172,219],"manage":[16],"them.":[17],"This":[18],"leads":[19],"problems":[21,136],"both:":[23],"have":[26],"low":[27],"utilization":[28],"when":[29,112],"the":[30,99,127,142,159,174],"traffic":[31],"load":[32],"is":[33],"low;":[34],"jobs":[36,124,198],"experience":[38],"long":[39],"queuing":[40,209],"due":[41],"a":[43,50,77,178,185],"lack":[44],"of":[45,144,168,194],"resources.":[46],"We":[47],"introduce":[48],"Lyra,":[49],"new":[51,93],"cluster":[52,96,215],"scheduler":[53],"address":[55],"these":[56,134],"problems.":[57],"Lyra":[58,132,201],"introduces":[59,141],"capacity":[60],"loaning":[61,88],"loan":[63],"idle":[64],"servers":[66,101],"jobs.":[69],"It":[70,140,155],"further":[71,156],"exploits":[72],"elastic":[73,90,123,170],"scaling":[74,91],"that":[75,200],"scales":[76],"job's":[79],"resource":[80],"allocation":[81],"better":[83],"utilize":[84],"loaned":[85,100],"servers.":[86],"Capacity":[87],"create":[92],"challenges":[94],"management.":[97],"When":[98],"need":[102,107,118],"be":[104],"returned,":[105],"we":[106,117],"minimize":[109,126],"job":[110,128,171],"preemptions;":[111],"more":[113],"GPUs":[114],"become":[115],"available,":[116],"allocate":[120],"them":[121],"completion":[129],"time":[130,210],"(JCT).":[131],"addresses":[133],"combinatorial":[135],"with":[137,191],"principled":[138],"heuristics.":[139],"notion":[143],"server":[145,153],"preemption":[146],"cost,":[147],"which":[148],"it":[149],"greedily":[150],"reduces":[151],"during":[152],"reclaiming.":[154],"relies":[157],"on":[158,184],"JCT":[160],"reduction":[161],"value":[162],"defined":[163],"each":[165],"additional":[166],"worker":[167],"an":[169],"solve":[173],"scheduling":[175],"problem":[176],"as":[177],"multiple-choice":[179],"knapsack":[180],"problem.":[181],"Prototype":[182],"implementation":[183],"64-GPU":[186],"testbed":[187],"large-scale":[189],"simulation":[190],"15-day":[192],"traces":[193],"over":[195],"50,000":[196],"production":[197],"show":[199],"brings":[202],"1.53x":[203],"1.48x":[205],"reductions":[206],"in":[207],"average":[208],"JCT,":[212],"improves":[214],"usage":[216],"by":[217],"up":[218],"25%.":[220]},"counts_by_year":[{"year":2026,"cited_by_count":6},{"year":2025,"cited_by_count":30},{"year":2024,"cited_by_count":14},{"year":2023,"cited_by_count":1}],"updated_date":"2026-07-31T08:31:51.225901","created_date":"2023-05-07T00:00:00"}
