{"id":"https://openalex.org/W4414855954","doi":"https://doi.org/10.1145/3801489.3806850","title":"Cache Your Prompt When It's Green - Carbon-aware Caching for Large Language Model Serving","display_name":"Cache Your Prompt When It's Green - Carbon-aware Caching for Large Language Model Serving","publication_year":2026,"publication_date":"2026-05-26","ids":{"openalex":"https://openalex.org/W4414855954","doi":"https://doi.org/10.1145/3801489.3806850"},"language":"en","primary_location":{"id":"doi:10.1145/3801489.3806850","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3801489.3806850","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Abstracts of the 2026 ACM SIGMETRICS International Conference on Measurement and Modeling of Computer Systems","raw_type":"proceedings-article"},"type":"article","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.1145/3801489.3806850","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5024567536","display_name":"Yanfei Tian","orcid":"https://orcid.org/0000-0001-7645-6918"},"institutions":[{"id":"https://openalex.org/I151746483","display_name":"University of Waterloo","ror":"https://ror.org/01aff2v68","country_code":"CA","type":"education","lineage":["https://openalex.org/I151746483"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Yuyang Tian","raw_affiliation_strings":["University of Waterloo, Waterloo, Canada"],"raw_orcid":"https://orcid.org/0009-0006-5438-1795","affiliations":[{"raw_affiliation_string":"University of Waterloo, Waterloo, Canada","institution_ids":["https://openalex.org/I151746483"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Desen Sun","orcid":"https://orcid.org/0000-0001-8630-7959"},"institutions":[{"id":"https://openalex.org/I151746483","display_name":"University of Waterloo","ror":"https://ror.org/01aff2v68","country_code":"CA","type":"education","lineage":["https://openalex.org/I151746483"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Desen Sun","raw_affiliation_strings":["University of Waterloo, Waterloo, Canada"],"raw_orcid":"https://orcid.org/0000-0001-8630-7959","affiliations":[{"raw_affiliation_string":"University of Waterloo, Waterloo, Canada","institution_ids":["https://openalex.org/I151746483"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5081538711","display_name":"Yi Ding","orcid":"https://orcid.org/0000-0003-4389-5636"},"institutions":[{"id":"https://openalex.org/I219193219","display_name":"Purdue University West Lafayette","ror":"https://ror.org/02dqehb95","country_code":"US","type":"education","lineage":["https://openalex.org/I219193219"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Yi Ding","raw_affiliation_strings":["Purdue University, West Lafayette, USA"],"raw_orcid":"https://orcid.org/0000-0003-2757-9182","affiliations":[{"raw_affiliation_string":"Purdue University, West Lafayette, USA","institution_ids":["https://openalex.org/I219193219"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5101571781","display_name":"Sihang Liu","orcid":"https://orcid.org/0000-0001-5711-8462"},"institutions":[{"id":"https://openalex.org/I151746483","display_name":"University of Waterloo","ror":"https://ror.org/01aff2v68","country_code":"CA","type":"education","lineage":["https://openalex.org/I151746483"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Sihang Liu","raw_affiliation_strings":["University of Waterloo, Waterloo, Canada"],"raw_orcid":"https://orcid.org/0000-0001-9706-6177","affiliations":[{"raw_affiliation_string":"University of Waterloo, Waterloo, Canada","institution_ids":["https://openalex.org/I151746483"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.00707747,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"94","last_page":"96"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11181","display_name":"Advanced Data Storage Technologies","score":0.9944000244140625,"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"}},"topics":[{"id":"https://openalex.org/T11181","display_name":"Advanced Data Storage Technologies","score":0.9944000244140625,"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":0.9828000068664551,"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/T11478","display_name":"Caching and Content Delivery","score":0.9800999760627747,"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/cache","display_name":"Cache","score":0.6583999991416931},{"id":"https://openalex.org/keywords/latency","display_name":"Latency (audio)","score":0.5318999886512756},{"id":"https://openalex.org/keywords/carbon-fibers","display_name":"Carbon fibers","score":0.4821000099182129},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.46140000224113464},{"id":"https://openalex.org/keywords/carbon-flux","display_name":"Carbon flux","score":0.438400000333786},{"id":"https://openalex.org/keywords/resource","display_name":"Resource (disambiguation)","score":0.3903000056743622},{"id":"https://openalex.org/keywords/reduction","display_name":"Reduction (mathematics)","score":0.3734999895095825}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7361000180244446},{"id":"https://openalex.org/C115537543","wikidata":"https://www.wikidata.org/wiki/Q165596","display_name":"Cache","level":2,"score":0.6583999991416931},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.5318999886512756},{"id":"https://openalex.org/C140205800","wikidata":"https://www.wikidata.org/wiki/Q5860","display_name":"Carbon fibers","level":3,"score":0.4821000099182129},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.46140000224113464},{"id":"https://openalex.org/C2993678126","wikidata":"https://www.wikidata.org/wiki/Q167751","display_name":"Carbon flux","level":3,"score":0.438400000333786},{"id":"https://openalex.org/C206345919","wikidata":"https://www.wikidata.org/wiki/Q20380951","display_name":"Resource (disambiguation)","level":2,"score":0.3903000056743622},{"id":"https://openalex.org/C111335779","wikidata":"https://www.wikidata.org/wiki/Q3454686","display_name":"Reduction (mathematics)","level":2,"score":0.3734999895095825},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.36309999227523804},{"id":"https://openalex.org/C192209626","wikidata":"https://www.wikidata.org/wiki/Q190909","display_name":"Focus (optics)","level":2,"score":0.3578999936580658},{"id":"https://openalex.org/C29202148","wikidata":"https://www.wikidata.org/wiki/Q287260","display_name":"Resource allocation","level":2,"score":0.3474999964237213},{"id":"https://openalex.org/C2780609101","wikidata":"https://www.wikidata.org/wiki/Q17156588","display_name":"Resource management (computing)","level":2,"score":0.31610000133514404},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.287200003862381},{"id":"https://openalex.org/C22607594","wikidata":"https://www.wikidata.org/wiki/Q5375150","display_name":"Enabling","level":2,"score":0.28369998931884766},{"id":"https://openalex.org/C168031717","wikidata":"https://www.wikidata.org/wiki/Q1530280","display_name":"Balance (ability)","level":2,"score":0.2590999901294708},{"id":"https://openalex.org/C100609095","wikidata":"https://www.wikidata.org/wiki/Q1335050","display_name":"Embodied cognition","level":2,"score":0.25429999828338623},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.25130000710487366}],"mesh":[],"locations_count":4,"locations":[{"id":"doi:10.1145/3801489.3806850","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3801489.3806850","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Abstracts of the 2026 ACM SIGMETRICS International Conference on Measurement and Modeling of Computer Systems","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2505.23970","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2505.23970","pdf_url":"https://arxiv.org/pdf/2505.23970","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"pmh:oai:arXiv.org:2505.23970","is_oa":true,"landing_page_url":"https://arxiv.org/abs/2505.23970","pdf_url":"https://arxiv.org/pdf/2505.23970","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2505.23970","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2505.23970","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.1145/3801489.3806850","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3801489.3806850","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Abstracts of the 2026 ACM SIGMETRICS International Conference on Measurement and Modeling of Computer Systems","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G1179075519","display_name":"Conference: DESC: Type III: A Holistic AI Computing Framework: Incorporating the Water and Biodiversity Dimensions of Sustainability","funder_award_id":"2413870","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"},{"id":"https://openalex.org/F4320334593","display_name":"Natural Sciences and Engineering Research Council of Canada","ror":"https://ror.org/01h531d29"}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"As":[0,70],"large":[1],"language":[2],"models":[3],"(LLMs)":[4],"become":[5],"widely":[6],"used,":[7],"their":[8],"environmental":[9],"impact":[10],"\u2014":[11,15],"especially":[12],"carbon":[13,25,51,65,75,106,123,138],"emission":[14,107,124],"has":[16],"attracted":[17],"more":[18],"attention.":[19],"Prior":[20],"studies":[21],"focus":[22],"on":[23],"compute-related":[24],"emissions.":[26],"In":[27],"this":[28,57,82],"paper,":[29],"we":[30,84],"find":[31],"that":[32,92,133],"storage":[33,77],"is":[34],"another":[35],"key":[36],"contributor.":[37],"LLM":[38,99],"caching,":[39],"which":[40],"saves":[41],"and":[42,108,122],"reuses":[43],"KV":[44],"caches":[45],"for":[46,98,163],"repeated":[47],"context,":[48],"reduces":[49],"operational":[50],"by":[52],"avoiding":[53],"redundant":[54],"computation.":[55],"However,":[56],"benefit":[58],"comes":[59],"at":[60],"the":[61,73,103,112,118,148],"cost":[62],"of":[63,76,140,167],"embodied":[64,74],"from":[66,129],"high-capacity,":[67],"high-speed":[68],"SSDs.":[69],"LLMs":[71],"scale,":[72],"grows":[78],"significantly.":[79],"To":[80],"address":[81],"tradeoff,":[83],"present":[85],"GreenCache,":[86],"a":[87],"carbon-aware":[88],"cache":[89],"management":[90],"framework":[91],"dynamically":[93],"derives":[94],"resource":[95,113],"allocation":[96],"plans":[97],"serving.":[100],"GreenCache":[101,134],"analyzes":[102],"correlation":[104],"between":[105,120],"SLO":[109,121],"satisfaction,":[110],"reconfiguring":[111],"over":[114],"time":[115],"to":[116,155],"keep":[117],"balance":[119],"under":[125],"dynamic":[126],"workloads.":[127],"Evaluations":[128],"real":[130],"traces":[131],"demonstrate":[132],"achieves":[135],"an":[136],"average":[137],"reduction":[139],"15.1":[141],"%":[142,166],"when":[143],"serving":[144],"Llama-3":[145],"70B":[146],"in":[147],"FR":[149],"grid,":[150],"with":[151],"reductions":[152],"reaching":[153],"up":[154],"25.3":[156],"%,":[157],"while":[158],"staying":[159],"within":[160],"latency":[161],"constraints":[162],">":[164],"90":[165],"requests.":[168]},"counts_by_year":[],"updated_date":"2026-05-28T06:12:49.907903","created_date":"2025-10-10T00:00:00"}