{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:LK57G3W6DX34HRMAJ3LDND7PP7","short_pith_number":"pith:LK57G3W6","schema_version":"1.0","canonical_sha256":"5abbf36ede1df7c3c5804ed6368fef7fccffc02e846d0167cb3b721e56a5533f","source":{"kind":"arxiv","id":"2504.18154","version":1},"attestation_state":"computed","paper":{"title":"EcoServe: Enabling Cost-effective LLM Serving with Proactive Intra- and Inter-Instance Orchestration","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.DC","authors_text":"Hongbin Zhang, Jiangsu Du, Kaiyi Wu, Taosheng Wei, Yutong Lu, Zhenyi Zheng, Zhiguang Chen","submitted_at":"2025-04-25T08:06:22Z","abstract_excerpt":"Existing LLM serving strategies can be categorized based on whether prefill and decode phases are disaggregated: non-disaggregated (NoDG) or fully disaggregated (FuDG). However, the NoDG strategy leads to strong prefill-decode interference and the FuDG strategy highly relies on high-performance interconnects, making them less cost-effective.\n  We introduce EcoServe, a system that enables cost-effective LLM serving on clusters with commodity interconnects. EcoServe is built on the partially disaggregated (PaDG) strategy, applying temporal disaggregation and rolling activation for proactive intr"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2504.18154","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2025-04-25T08:06:22Z","cross_cats_sorted":[],"title_canon_sha256":"487d6e0310de58e98f9af21efc1786c22554830b11286953e5af4f3b8aab5643","abstract_canon_sha256":"6110d1894df0054e9375864fe3716777214163987a26d440118f3eef799fd914"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:54:02.874465Z","signature_b64":"QTIO7DE9CLANuJzuq3RCt+TWsm2MBaM2c1L66UduYxSATGaFTQGRL4cAJMlr5uIPawHQVK2YqOeHh6pe4BR3DA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5abbf36ede1df7c3c5804ed6368fef7fccffc02e846d0167cb3b721e56a5533f","last_reissued_at":"2026-07-05T10:54:02.873978Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:54:02.873978Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"EcoServe: Enabling Cost-effective LLM Serving with Proactive Intra- and Inter-Instance Orchestration","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.DC","authors_text":"Hongbin Zhang, Jiangsu Du, Kaiyi Wu, Taosheng Wei, Yutong Lu, Zhenyi Zheng, Zhiguang Chen","submitted_at":"2025-04-25T08:06:22Z","abstract_excerpt":"Existing LLM serving strategies can be categorized based on whether prefill and decode phases are disaggregated: non-disaggregated (NoDG) or fully disaggregated (FuDG). However, the NoDG strategy leads to strong prefill-decode interference and the FuDG strategy highly relies on high-performance interconnects, making them less cost-effective.\n  We introduce EcoServe, a system that enables cost-effective LLM serving on clusters with commodity interconnects. EcoServe is built on the partially disaggregated (PaDG) strategy, applying temporal disaggregation and rolling activation for proactive intr"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.18154","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2504.18154/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2504.18154","created_at":"2026-07-05T10:54:02.874034+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.18154v1","created_at":"2026-07-05T10:54:02.874034+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.18154","created_at":"2026-07-05T10:54:02.874034+00:00"},{"alias_kind":"pith_short_12","alias_value":"LK57G3W6DX34","created_at":"2026-07-05T10:54:02.874034+00:00"},{"alias_kind":"pith_short_16","alias_value":"LK57G3W6DX34HRMA","created_at":"2026-07-05T10:54:02.874034+00:00"},{"alias_kind":"pith_short_8","alias_value":"LK57G3W6","created_at":"2026-07-05T10:54:02.874034+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.05933","citing_title":"Beyond Greedy Chunking: SLO-Aware Sliding-Window Scheduling for LLM Inference","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2605.02189","citing_title":"PipeMax: Enhancing Offline LLM Inference on Commodity GPU Servers","ref_index":11,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LK57G3W6DX34HRMAJ3LDND7PP7","json":"https://pith.science/pith/LK57G3W6DX34HRMAJ3LDND7PP7.json","graph_json":"https://pith.science/api/pith-number/LK57G3W6DX34HRMAJ3LDND7PP7/graph.json","events_json":"https://pith.science/api/pith-number/LK57G3W6DX34HRMAJ3LDND7PP7/events.json","paper":"https://pith.science/paper/LK57G3W6"},"agent_actions":{"view_html":"https://pith.science/pith/LK57G3W6DX34HRMAJ3LDND7PP7","download_json":"https://pith.science/pith/LK57G3W6DX34HRMAJ3LDND7PP7.json","view_paper":"https://pith.science/paper/LK57G3W6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.18154&json=true","fetch_graph":"https://pith.science/api/pith-number/LK57G3W6DX34HRMAJ3LDND7PP7/graph.json","fetch_events":"https://pith.science/api/pith-number/LK57G3W6DX34HRMAJ3LDND7PP7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LK57G3W6DX34HRMAJ3LDND7PP7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LK57G3W6DX34HRMAJ3LDND7PP7/action/storage_attestation","attest_author":"https://pith.science/pith/LK57G3W6DX34HRMAJ3LDND7PP7/action/author_attestation","sign_citation":"https://pith.science/pith/LK57G3W6DX34HRMAJ3LDND7PP7/action/citation_signature","submit_replication":"https://pith.science/pith/LK57G3W6DX34HRMAJ3LDND7PP7/action/replication_record"}},"created_at":"2026-07-05T10:54:02.874034+00:00","updated_at":"2026-07-05T10:54:02.874034+00:00"}