{"id":"https://openalex.org/W4411403155","doi":"https://doi.org/10.1145/3725274","title":"CARINA: An Efficient CXL-Oriented Embedding Serving System for Recommendation Models","display_name":"CARINA: An Efficient CXL-Oriented Embedding Serving System for Recommendation Models","publication_year":2025,"publication_date":"2025-06-17","ids":{"openalex":"https://openalex.org/W4411403155","doi":"https://doi.org/10.1145/3725274"},"language":"en","primary_location":{"id":"doi:10.1145/3725274","doi":"https://doi.org/10.1145/3725274","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3725274","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3725274","source":{"id":"https://openalex.org/S4387289859","display_name":"Proceedings of the ACM on Management of Data","issn_l":"2836-6573","issn":["2836-6573"],"is_oa":false,"is_in_doaj":false,"is_core":true,"listed_in":["cwts-core","norway-1","vabb-shw"],"host_organization":"https://openalex.org/P4310319798","host_organization_name":"Association for Computing Machinery","host_organization_lineage":["https://openalex.org/P4310319798"],"host_organization_lineage_names":["Association for Computing Machinery"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the ACM on Management of Data","raw_type":"journal-article"},"volume":null,"type":"article","is_traditional":true,"indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://dl.acm.org/doi/pdf/10.1145/3725274","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5102854732","display_name":"Peiqi Yin","orcid":"https://orcid.org/0009-0008-4081-913X","observed_orcids":["https://orcid.org/0009-0008-4081-913X"]},"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":"Peiqi Yin","raw_affiliation_strings":["The Chinese University of Hong Kong, Hong Kong SAR, China"],"raw_orcid":"https://orcid.org/0009-0008-4081-913X","affiliations":[{"raw_affiliation_string":"The Chinese University of Hong Kong, Hong Kong SAR, China","institution_ids":["https://openalex.org/I177725633"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5045851375","display_name":"Qihui Zhou","orcid":"https://orcid.org/0000-0002-8927-3325","observed_orcids":["https://orcid.org/0000-0002-8927-3325"]},"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":"Qihui Zhou","raw_affiliation_strings":["The Chinese University of Hong Kong, Hong Kong SAR, China"],"raw_orcid":"https://orcid.org/0000-0002-8927-3325","affiliations":[{"raw_affiliation_string":"The Chinese University of Hong Kong, Hong Kong SAR, China","institution_ids":["https://openalex.org/I177725633"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100367774","display_name":"Xiao Yan","orcid":"https://orcid.org/0000-0002-2122-915X","observed_orcids":["https://orcid.org/0000-0002-2122-915X","https://orcid.org/0000-0002-2563-083X"]},"institutions":[{"id":"https://openalex.org/I4405270501","display_name":"Centre for Perceptual and Interactive Intelligence","ror":"https://ror.org/00tdmgp49","country_code":"HK","type":"facility","lineage":["https://openalex.org/I4405270501"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Xiao Yan","raw_affiliation_strings":["Centre for Perceptual and Interactive Intelligence (CPII), Hong Kong SAR, China"],"raw_orcid":"https://orcid.org/0000-0002-2122-915X","affiliations":[{"raw_affiliation_string":"Centre for Perceptual and Interactive Intelligence (CPII), Hong Kong SAR, China","institution_ids":["https://openalex.org/I4405270501"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148327535","display_name":"Chao Wang","orcid":"https://orcid.org/0009-0003-2162-5675","observed_orcids":["https://orcid.org/0009-0003-2162-5675"]},"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":"Chao Wang","raw_affiliation_strings":["The Chinese University of Hong Kong, Hong Kong SAR, China"],"raw_orcid":"https://orcid.org/0009-0003-2162-5675","affiliations":[{"raw_affiliation_string":"The Chinese University of Hong Kong, Hong Kong SAR, China","institution_ids":["https://openalex.org/I177725633"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5011148671","display_name":"Eric Lo","orcid":"https://orcid.org/0000-0003-2679-3945","observed_orcids":["https://orcid.org/0000-0003-2679-3945"]},"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":"Eric Lo","raw_affiliation_strings":["The Chinese University of Hong Kong, Hong Kong SAR, China"],"raw_orcid":"https://orcid.org/0000-0003-2679-3945","affiliations":[{"raw_affiliation_string":"The Chinese University of Hong Kong, Hong Kong SAR, China","institution_ids":["https://openalex.org/I177725633"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5085580462","display_name":"Changji Li","orcid":"https://orcid.org/0000-0003-2768-1224","observed_orcids":["https://orcid.org/0000-0003-2768-1224"]},"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":"Changji Li","raw_affiliation_strings":["The Chinese University of Hong Kong, Hong Kong SAR, China"],"raw_orcid":"https://orcid.org/0000-0003-2768-1224","affiliations":[{"raw_affiliation_string":"The Chinese University of Hong Kong, Hong Kong SAR, China","institution_ids":["https://openalex.org/I177725633"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5021814788","display_name":"Lan Lu","orcid":"https://orcid.org/0009-0003-6492-5221","observed_orcids":["https://orcid.org/0009-0003-6492-5221"]},"institutions":[{"id":"https://openalex.org/I79576946","display_name":"University of Pennsylvania","ror":"https://ror.org/00b30xv10","country_code":"US","type":"education","lineage":["https://openalex.org/I79576946"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Lan Lu","raw_affiliation_strings":["University of Pennsylvania, Philadelphia, PA, USA"],"raw_orcid":"https://orcid.org/0009-0003-6492-5221","affiliations":[{"raw_affiliation_string":"University of Pennsylvania, Philadelphia, PA, USA","institution_ids":["https://openalex.org/I79576946"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5088804666","display_name":"Hua Tao Fan","orcid":"https://orcid.org/0000-0001-6230-0445","observed_orcids":["https://orcid.org/0000-0001-6230-0445"]},"institutions":[{"id":"https://openalex.org/I45928872","display_name":"Alibaba Group (China)","ror":"https://ror.org/00k642b80","country_code":"CN","type":"company","lineage":["https://openalex.org/I45928872"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Hua Fan","raw_affiliation_strings":["Alibaba Cloud, Hangzhou, China"],"raw_orcid":"https://orcid.org/0000-0001-6230-0445","affiliations":[{"raw_affiliation_string":"Alibaba Cloud, Hangzhou, China","institution_ids":["https://openalex.org/I45928872"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101741069","display_name":"Wenchao Zhou","orcid":"https://orcid.org/0009-0002-2689-6020","observed_orcids":["https://orcid.org/0009-0002-2689-6020"]},"institutions":[{"id":"https://openalex.org/I45928872","display_name":"Alibaba Group (China)","ror":"https://ror.org/00k642b80","country_code":"CN","type":"company","lineage":["https://openalex.org/I45928872"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wenchao Zhou","raw_affiliation_strings":["Alibaba Cloud, Hangzhou, China"],"raw_orcid":"https://orcid.org/0009-0002-2689-6020","affiliations":[{"raw_affiliation_string":"Alibaba Cloud, Hangzhou, China","institution_ids":["https://openalex.org/I45928872"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102930563","display_name":"Ming-Chang Yang","orcid":"https://orcid.org/0000-0002-4029-757X","observed_orcids":["https://orcid.org/0000-0002-4029-757X"]},"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":"Ming-Chang Yang","raw_affiliation_strings":["The Chinese University of Hong Kong, Hong Kong SAR, China"],"raw_orcid":"https://orcid.org/0000-0002-4029-757X","affiliations":[{"raw_affiliation_string":"The Chinese University of Hong Kong, Hong Kong SAR, China","institution_ids":["https://openalex.org/I177725633"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5016082884","display_name":"James Sheung-Chak Cheng","orcid":"https://orcid.org/0000-0001-6313-6288","observed_orcids":["https://orcid.org/0000-0001-6313-6288","https://orcid.org/0009-0008-7820-2506"]},"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":"James Cheng","raw_affiliation_strings":["The Chinese University of Hong Kong, Hong Kong SAR, China"],"raw_orcid":"https://orcid.org/0000-0001-6313-6288","affiliations":[{"raw_affiliation_string":"The Chinese University of Hong Kong, Hong Kong SAR, China","institution_ids":["https://openalex.org/I177725633"]}]}],"institutions":[],"countries_distinct_count":3,"institutions_distinct_count":4,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":1.2047,"has_fulltext":true,"cited_by_count":2,"citation_normalized_percentile":{"value":0.81701195,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":95,"max":96},"biblio":{"volume":"3","issue":"3","first_page":"1","last_page":"29"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10203","display_name":"Recommender Systems and Techniques","score":0.37380000948905945,"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/T10203","display_name":"Recommender Systems and Techniques","score":0.37380000948905945,"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":0.36570000648498535,"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/T10317","display_name":"Advanced Database Systems and Queries","score":0.14589999616146088,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/cxl","display_name":"CXL","score":0.765999972820282},{"id":"https://openalex.org/keywords/heterogeneous-memory","display_name":"heterogeneous memory","score":0.8190000057220459},{"id":"https://openalex.org/keywords/numa","display_name":"NUMA","score":0.550000011920929},{"id":"https://openalex.org/keywords/remote-memory","display_name":"remote memory","score":0.4970000088214874},{"id":"https://openalex.org/keywords/request-latency","display_name":"request latency","score":0.39500001072883606}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8455512523651123},{"id":"https://openalex.org/C7366592","wikidata":"https://www.wikidata.org/wiki/Q1255620","display_name":"Dram","level":2,"score":0.5735286474227905},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.525033712387085},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.45743778347969055},{"id":"https://openalex.org/C2776257435","wikidata":"https://www.wikidata.org/wiki/Q1576430","display_name":"Bandwidth (computing)","level":2,"score":0.41323432326316833},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.29065829515457153},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.2772331237792969},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.2354939877986908}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3725274","doi":"https://doi.org/10.1145/3725274","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3725274","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3725274","source":{"id":"https://openalex.org/S4387289859","display_name":"Proceedings of the ACM on Management of Data","issn_l":"2836-6573","issn":["2836-6573"],"is_oa":false,"is_in_doaj":false,"is_core":true,"listed_in":["cwts-core","norway-1","vabb-shw"],"host_organization":"https://openalex.org/P4310319798","host_organization_name":"Association for Computing Machinery","host_organization_lineage":["https://openalex.org/P4310319798"],"host_organization_lineage_names":["Association for Computing Machinery"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the ACM on Management of Data","raw_type":"journal-article"}],"best_oa_location":{"id":"doi:10.1145/3725274","doi":"https://doi.org/10.1145/3725274","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3725274","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3725274","source":{"id":"https://openalex.org/S4387289859","display_name":"Proceedings of the ACM on Management of Data","issn_l":"2836-6573","issn":["2836-6573"],"is_oa":false,"is_in_doaj":false,"is_core":true,"listed_in":["cwts-core","norway-1","vabb-shw"],"host_organization":"https://openalex.org/P4310319798","host_organization_name":"Association for Computing Machinery","host_organization_lineage":["https://openalex.org/P4310319798"],"host_organization_lineage_names":["Association for Computing Machinery"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the ACM on Management of Data","raw_type":"journal-article"},"sustainable_development_goals":[{"display_name":"Industry, innovation and infrastructure","score":0.428600013256073,"id":"https://metadata.un.org/sdg/9"}],"study_designs":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320321920","display_name":"Innovation and Technology Commission","ror":"https://ror.org/04vf9tr09"},{"id":"https://openalex.org/F4320323537","display_name":"Hong Kong University of Science and Technology","ror":"https://ror.org/00q4vv597"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4411403155.pdf","grobid_xml":"https://content.openalex.org/works/W4411403155.grobid-xml"},"referenced_works_count":72,"referenced_works":["https://openalex.org/W1986398087","https://openalex.org/W2133386065","https://openalex.org/W2475334473","https://openalex.org/W2489773810","https://openalex.org/W2591921973","https://openalex.org/W2604662567","https://openalex.org/W2793768763","https://openalex.org/W2794568143","https://openalex.org/W2963469388","https://openalex.org/W3016842236","https://openalex.org/W3043023836","https://openalex.org/W3043433718","https://openalex.org/W3080555959","https://openalex.org/W3095776306","https://openalex.org/W3118280765","https://openalex.org/W3130104841","https://openalex.org/W3153887672","https://openalex.org/W3155243801","https://openalex.org/W3163805047","https://openalex.org/W3166125679","https://openalex.org/W3166863226","https://openalex.org/W3187188899","https://openalex.org/W3200211247","https://openalex.org/W4214658871","https://openalex.org/W4220660812","https://openalex.org/W4220884018","https://openalex.org/W4221150565","https://openalex.org/W4226012237","https://openalex.org/W4231769592","https://openalex.org/W4280567565","https://openalex.org/W4280642915","https://openalex.org/W4282555269","https://openalex.org/W4290928031","https://openalex.org/W4296591817","https://openalex.org/W4296591836","https://openalex.org/W4310744178","https://openalex.org/W4318541517","https://openalex.org/W4318541545","https://openalex.org/W4318623169","https://openalex.org/W4318767786","https://openalex.org/W4319866214","https://openalex.org/W4321466207","https://openalex.org/W4323343906","https://openalex.org/W4327930439","https://openalex.org/W4361194512","https://openalex.org/W4377131073","https://openalex.org/W4380881110","https://openalex.org/W4381327166","https://openalex.org/W4381329135","https://openalex.org/W4381329179","https://openalex.org/W4383749443","https://openalex.org/W4385270365","https://openalex.org/W4385567855","https://openalex.org/W4386568791","https://openalex.org/W4387302703","https://openalex.org/W4387321090","https://openalex.org/W4388581411","https://openalex.org/W4388644533","https://openalex.org/W4391250471","https://openalex.org/W4393145094","https://openalex.org/W4393638414","https://openalex.org/W4394946262","https://openalex.org/W4396843640","https://openalex.org/W4399174662","https://openalex.org/W4399304239","https://openalex.org/W4401211935","https://openalex.org/W4401212200","https://openalex.org/W4402969637","https://openalex.org/W4404954501","https://openalex.org/W4413014932","https://openalex.org/W4414371540","https://openalex.org/W6601101692"],"related_works":["https://openalex.org/W4401212200","https://openalex.org/W7214646183","https://openalex.org/W7163559621","https://openalex.org/W4361194512","https://openalex.org/W7164347459","https://openalex.org/W4416800428","https://openalex.org/W4413968273","https://openalex.org/W4319866214","https://openalex.org/W4391250471","https://openalex.org/W4376122538"],"abstract_inverted_index":{"Embedding-based":[0],"recommendation":[1],"models":[2],"(ERMs)":[3],"require":[4],"large":[5],"memory":[6,34,38,69,74,98,111,115],"to":[7,17,158,166],"host":[8],"huge":[9],"embedding":[10,132],"tables":[11,133],"and":[12,62,127,154,177,190,195],"involve":[13],"massive":[14],"data":[15],"traffic":[16],"read":[18],"the":[19,53,70,84,110,129,135,156,160],"embeddings.":[20],"As":[21],"a":[22,182],"new":[23],"interconnect,":[24],"CXL":[25,48,56,68,80,100,164,175],"suits":[26],"ERMs":[27],"since":[28],"it":[29,180],"can":[30,63],"scale":[31],"up":[32],"single-machine":[33],"with":[35,99,121],"performant":[36],"remote":[37],"devices.":[39],"However,":[40],"directly":[41],"running":[42],"DRAM-based":[43],"ERM":[44,94,149],"serving":[45,95],"systems":[46],"on":[47,96,125,134],"yields":[49],"poor":[50],"performance":[51],"because":[52],"bandwidth":[54,104,165],"of":[55,131,148,163,188],"is":[57],"notably":[58],"lower":[59],"than":[60],"DRAM":[61,126],"be":[64],"easily":[65],"saturated,":[66],"making":[67],"bottleneck.":[71],"The":[72],"non-uniform":[73],"access":[75,112,123],"(NUMA)":[76],"architecture":[77],"in":[78,192],"modern":[79],"servers":[81],"further":[82],"decreased":[83],"system":[85,193],"performance.":[86],"In":[87,106],"this":[88],"paper,":[89],"we":[90],"design":[91],"Carina":[92,108,139,172],"for":[93],"heterogeneous":[97],"by":[101,117,185],"considering":[102],"such":[103],"asymmetry.":[105],"particular,":[107],"balances":[109],"from":[113],"different":[114],"devices":[116,176],"storing":[118],"hot":[119],"embeddings":[120],"high":[122],"frequencies":[124],"specifying":[128],"placement":[130],"NUMA":[136],"nodes.":[137],"Moreover,":[138],"adopts":[140],"bandwidth-aware":[141],"task":[142],"execution,":[143],"which":[144],"decomposes":[145],"each":[146],"batch":[147],"requests":[150],"into":[151],"fine-grained":[152],"tasks":[153,157],"schedules":[155],"control":[159],"real-time":[161],"utilization":[162],"avoid":[167],"instantaneous":[168],"saturation.":[169],"We":[170],"evaluate":[171],"under":[173],"real":[174],"find":[178],"that":[179],"outperforms":[181],"CXL-oblivious":[183],"baseline":[184],"an":[186],"average":[187],"5.38x":[189],"4.04x":[191],"throughput":[194],"request":[196],"latency,":[197],"respectively.":[198]},"counts_by_year":[{"year":2026,"cited_by_count":2}],"updated_date":"2026-11-04T10:54:05.112534","created_date":"2025-10-10T00:00:00"}
