{"id":"https://openalex.org/W7123486245","doi":"https://doi.org/10.1145/3772052.3772264","title":"C <scp>auchy</scp> : A Cost-Efficient LLM Serving System through Adaptive Heterogeneous Deployment","display_name":"C <scp>auchy</scp> : A Cost-Efficient LLM Serving System through Adaptive Heterogeneous Deployment","publication_year":2025,"publication_date":"2025-11-19","ids":{"openalex":"https://openalex.org/W7123486245","doi":"https://doi.org/10.1145/3772052.3772264"},"language":null,"primary_location":{"id":"doi:10.1145/3772052.3772264","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3772052.3772264","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2025 ACM Symposium on Cloud Computing","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Yihui Zhang","orcid":"https://orcid.org/0009-0009-5768-0063"},"institutions":[{"id":"https://openalex.org/I82880672","display_name":"Beihang University","ror":"https://ror.org/00wk2mp56","country_code":"CN","type":"education","lineage":["https://openalex.org/I82880672"]}],"countries":["CN"],"is_corresponding":true,"raw_author_name":"Yihui Zhang","raw_affiliation_strings":["Beihang University, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0009-5768-0063","affiliations":[{"raw_affiliation_string":"Beihang University, Beijing, China","institution_ids":["https://openalex.org/I82880672"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5122941251","display_name":"han Shen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Han Shen","raw_affiliation_strings":["Kuaishou Inc., Beijing, China"],"raw_orcid":"https://orcid.org/0009-0003-0517-5467","affiliations":[{"raw_affiliation_string":"Kuaishou Inc., Beijing, China","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5050796169","display_name":"Renyu Yang","orcid":"https://orcid.org/0000-0001-6334-4925"},"institutions":[{"id":"https://openalex.org/I82880672","display_name":"Beihang University","ror":"https://ror.org/00wk2mp56","country_code":"CN","type":"education","lineage":["https://openalex.org/I82880672"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Renyu Yang","raw_affiliation_strings":["Beihang University, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0001-6334-4925","affiliations":[{"raw_affiliation_string":"Beihang University, Beijing, China","institution_ids":["https://openalex.org/I82880672"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5122928573","display_name":"Di Tian","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Di Tian","raw_affiliation_strings":["Kuaishou Inc., Beijing, China"],"raw_orcid":"https://orcid.org/0009-0002-7618-6668","affiliations":[{"raw_affiliation_string":"Kuaishou Inc., Beijing, China","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5122940766","display_name":"Yuxi Luo","orcid":null},"institutions":[{"id":"https://openalex.org/I82880672","display_name":"Beihang University","ror":"https://ror.org/00wk2mp56","country_code":"CN","type":"education","lineage":["https://openalex.org/I82880672"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yuxi Luo","raw_affiliation_strings":["Beihang University, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0004-7584-1020","affiliations":[{"raw_affiliation_string":"Beihang University, Beijing, China","institution_ids":["https://openalex.org/I82880672"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5031176052","display_name":"Menghao Zhang","orcid":"https://orcid.org/0000-0001-5274-5512"},"institutions":[{"id":"https://openalex.org/I82880672","display_name":"Beihang University","ror":"https://ror.org/00wk2mp56","country_code":"CN","type":"education","lineage":["https://openalex.org/I82880672"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Menghao Zhang","raw_affiliation_strings":["Beihang University, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0001-5274-5512","affiliations":[{"raw_affiliation_string":"Beihang University, Beijing, China","institution_ids":["https://openalex.org/I82880672"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5122928190","display_name":"Li Li","orcid":null},"institutions":[{"id":"https://openalex.org/I82880672","display_name":"Beihang University","ror":"https://ror.org/00wk2mp56","country_code":"CN","type":"education","lineage":["https://openalex.org/I82880672"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Li Li","raw_affiliation_strings":["Beihang University, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0003-2990-1614","affiliations":[{"raw_affiliation_string":"Beihang University, Beijing, China","institution_ids":["https://openalex.org/I82880672"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5122977824","display_name":"Chunming Hu","orcid":null},"institutions":[{"id":"https://openalex.org/I82880672","display_name":"Beihang University","ror":"https://ror.org/00wk2mp56","country_code":"CN","type":"education","lineage":["https://openalex.org/I82880672"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Chunming Hu","raw_affiliation_strings":["Beihang University, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0003-3473-9703","affiliations":[{"raw_affiliation_string":"Beihang University, Beijing, China","institution_ids":["https://openalex.org/I82880672"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5066123431","display_name":"Tianyu Wo","orcid":"https://orcid.org/0000-0002-5331-3364"},"institutions":[{"id":"https://openalex.org/I82880672","display_name":"Beihang University","ror":"https://ror.org/00wk2mp56","country_code":"CN","type":"education","lineage":["https://openalex.org/I82880672"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Tianyu Wo","raw_affiliation_strings":["Beihang University, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0002-5331-3364","affiliations":[{"raw_affiliation_string":"Beihang University, Beijing, China","institution_ids":["https://openalex.org/I82880672"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5122942158","display_name":"Chengru Song","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chengru Song","raw_affiliation_strings":["Kuaishou Inc., Beijing, China"],"raw_orcid":"https://orcid.org/0009-0009-3826-8436","affiliations":[{"raw_affiliation_string":"Kuaishou Inc., Beijing, China","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5103963648","display_name":"Jin Ouyang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jin Ouyang","raw_affiliation_strings":["Kuaishou Inc., Beijing, China"],"raw_orcid":"https://orcid.org/0009-0003-6591-7717","affiliations":[{"raw_affiliation_string":"Kuaishou Inc., Beijing, China","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":11,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I82880672"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.78546884,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"881","last_page":"893"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T14347","display_name":"Big Data and Digital Economy","score":0.2712000012397766,"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/T14347","display_name":"Big Data and Digital Economy","score":0.2712000012397766,"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/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.17159999907016754,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/T10181","display_name":"Natural Language Processing Techniques","score":0.11580000072717667,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"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/scheduling","display_name":"Scheduling (production processes)","score":0.6536999940872192},{"id":"https://openalex.org/keywords/workload","display_name":"Workload","score":0.5756000280380249},{"id":"https://openalex.org/keywords/software-deployment","display_name":"Software deployment","score":0.5253999829292297},{"id":"https://openalex.org/keywords/quality-of-service","display_name":"Quality of service","score":0.4984999895095825},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.4821999967098236},{"id":"https://openalex.org/keywords/symmetric-multiprocessor-system","display_name":"Symmetric multiprocessor system","score":0.4047999978065491},{"id":"https://openalex.org/keywords/homogeneous","display_name":"Homogeneous","score":0.3723999857902527},{"id":"https://openalex.org/keywords/multi-core-processor","display_name":"Multi-core processor","score":0.36230000853538513}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8210999965667725},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.6536999940872192},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.638700008392334},{"id":"https://openalex.org/C2778476105","wikidata":"https://www.wikidata.org/wiki/Q628539","display_name":"Workload","level":2,"score":0.5756000280380249},{"id":"https://openalex.org/C105339364","wikidata":"https://www.wikidata.org/wiki/Q2297740","display_name":"Software deployment","level":2,"score":0.5253999829292297},{"id":"https://openalex.org/C5119721","wikidata":"https://www.wikidata.org/wiki/Q220501","display_name":"Quality of service","level":2,"score":0.4984999895095825},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.4821999967098236},{"id":"https://openalex.org/C172430144","wikidata":"https://www.wikidata.org/wiki/Q17111997","display_name":"Symmetric multiprocessor system","level":2,"score":0.4047999978065491},{"id":"https://openalex.org/C66882249","wikidata":"https://www.wikidata.org/wiki/Q169336","display_name":"Homogeneous","level":2,"score":0.3723999857902527},{"id":"https://openalex.org/C78766204","wikidata":"https://www.wikidata.org/wiki/Q555032","display_name":"Multi-core processor","level":2,"score":0.36230000853538513},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.35899999737739563},{"id":"https://openalex.org/C206345919","wikidata":"https://www.wikidata.org/wiki/Q20380951","display_name":"Resource (disambiguation)","level":2,"score":0.3431999981403351},{"id":"https://openalex.org/C79974875","wikidata":"https://www.wikidata.org/wiki/Q483639","display_name":"Cloud computing","level":2,"score":0.33660000562667847},{"id":"https://openalex.org/C107568181","wikidata":"https://www.wikidata.org/wiki/Q5319000","display_name":"Dynamic priority scheduling","level":3,"score":0.3172999918460846},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.3021000027656555},{"id":"https://openalex.org/C2777958785","wikidata":"https://www.wikidata.org/wiki/Q17120940","display_name":"Resource efficiency","level":2,"score":0.28760001063346863},{"id":"https://openalex.org/C29202148","wikidata":"https://www.wikidata.org/wiki/Q287260","display_name":"Resource allocation","level":2,"score":0.28529998660087585},{"id":"https://openalex.org/C177212765","wikidata":"https://www.wikidata.org/wiki/Q627335","display_name":"Workflow","level":2,"score":0.28029999136924744},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.2662000060081482},{"id":"https://openalex.org/C50630238","wikidata":"https://www.wikidata.org/wiki/Q971505","display_name":"General-purpose computing on graphics processing units","level":3,"score":0.2619999945163727},{"id":"https://openalex.org/C51332947","wikidata":"https://www.wikidata.org/wiki/Q1172305","display_name":"Shared resource","level":2,"score":0.2599000036716461},{"id":"https://openalex.org/C49344536","wikidata":"https://www.wikidata.org/wiki/Q726441","display_name":"Cauchy distribution","level":2,"score":0.25110000371932983}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3772052.3772264","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3772052.3772264","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2025 ACM Symposium on Cloud Computing","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G6618872119","display_name":null,"funder_award_id":"62402024","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G8792818388","display_name":null,"funder_award_id":"2024YFB4505901","funder_id":"https://openalex.org/F4320335777","funder_display_name":"National Key Research and Development Program of China"}],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"},{"id":"https://openalex.org/F4320335777","display_name":"National Key Research and Development Program of China","ror":null}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":3,"referenced_works":["https://openalex.org/W2606974598","https://openalex.org/W4387321091","https://openalex.org/W4401211704"],"related_works":[],"abstract_inverted_index":{"Recent":[0],"advances":[1],"in":[2,98,150,166],"large":[3],"language":[4],"models":[5],"(LLMs)":[6],"have":[7],"intensified":[8],"the":[9,66,77,130,148,151,170,183],"need":[10],"for":[11,94],"serving":[12,56],"LLMs":[13],"that":[14,58,158],"are":[15],"cost-efficient":[16,194],"and":[17,26,37,47,62,71,134,188],"QoS-guaranteed.":[18],"Existing":[19],"frameworks":[20],"often":[21],"co-locate":[22],"computationally":[23],"distinct":[24],"prefill":[25,61],"decode":[27,63],"instances":[28],"on":[29],"homogeneous":[30],"GPUs,":[31],"overlooking":[32],"their":[33,95],"unique":[34],"resource":[35,45],"demands":[36],"under-utilizing":[38],"heterogeneous":[39,69],"GPUs.":[40],"This":[41],"leads":[42],"to":[43,65,108,122,146,162,191],"suboptimal":[44],"utilization":[46],"increased":[48],"capital":[49],"expenditure.":[50],"We":[51],"present":[52],"Cauchy,":[53],"a":[54,86,104,135,163],"LLM":[55,115,195],"framework":[57],"adaptively":[59],"deploys":[60,103],"computation":[64],"most":[67],"suitable":[68],"GPUs":[70],"dynamically":[72],"schedules":[73],"user":[74,124],"requests.":[75,155],"At":[76],"core":[78],"of":[79,106,114,153,185],"Cauchy":[80,102,117,159],"is":[81,144],"choosing":[82],"proper":[83],"GPU":[84,88,92,132,140,189],"Combo,":[85],"conceptual":[87],"combination":[89],"encompassing":[90],"diverse":[91],"configurations,":[93],"cost":[96],"efficiency":[97,168],"running":[99],"prefill-decode":[100],"pairs.":[101],"set":[105],"combos":[107],"satisfy":[109],"QoS":[110],"requirements":[111],"(e.g.,":[112],"goodput)":[113],"inference.":[116],"further":[118],"employs":[119],"hierarchical":[120],"scheduling":[121,128],"handle":[123],"requests,":[125],"using":[126],"opportunistic":[127],"within":[129],"allocated":[131],"Combos":[133],"goodput-weighted":[136],"round-robin":[137],"policy":[138],"across":[139],"Combos.":[141],"Dynamic":[142],"autoscaling":[143],"used":[145],"stabilize":[147],"cost-efficiency":[149],"face":[152],"surging":[154],"Experiments":[156],"show":[157],"achieves":[160],"up":[161],"38.3%":[164],"improvement":[165],"Tokens/USD":[167],"over":[169],"state-of-the-art":[171],"baselines,":[172],"while":[173],"maintaining":[174],"strict":[175],"Service":[176],"Level":[177],"Objectives":[178],"(SLOs).":[179],"Our":[180],"work":[181],"highlights":[182],"importance":[184],"leveraging":[186],"workload":[187],"heterogeneity":[190],"achieve":[192],"superior":[193],"serving.":[196]},"counts_by_year":[],"updated_date":"2026-01-14T23:44:37.837170","created_date":"2026-01-14T00:00:00"}