{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:D6KX2MO3IYZGB5HWNZS5PCM5W3","short_pith_number":"pith:D6KX2MO3","schema_version":"1.0","canonical_sha256":"1f957d31db463260f4f66e65d7899db6e9ae18ae3c4fcff06c667dd561db483c","source":{"kind":"arxiv","id":"2508.19559","version":1},"attestation_state":"computed","paper":{"title":"Taming the Chaos: Coordinated Autoscaling for Heterogeneous and Disaggregated LLM Inference","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.DC","authors_text":"Chunlei Han, Huanle Han, Long Huang, Rongzhi Li, Ruogu Du, Shufan Liu, Sida Zhao, Yiwen Shao, Zefang Chu, Zherui Liu, Zuocheng Shi","submitted_at":"2025-08-27T04:22:02Z","abstract_excerpt":"Serving Large Language Models (LLMs) is a GPU-intensive task where traditional autoscalers fall short, particularly for modern Prefill-Decode (P/D) disaggregated architectures. This architectural shift, while powerful, introduces significant operational challenges, including inefficient use of heterogeneous hardware, network bottlenecks, and critical imbalances between prefill and decode stages. We introduce HeteroScale, a coordinated autoscaling framework that addresses the core challenges of P/D disaggregated serving. HeteroScale combines a topology-aware scheduler that adapts to heterogeneo"},"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":"2508.19559","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2025-08-27T04:22:02Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"eacb06064449ea851be690db00deac890de807115029b818200d895caabe1f5b","abstract_canon_sha256":"cb17d8051316eaa467b7f3dd9acaf759d8fdc0370b82437c15731b9ee9a94073"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:00:14.095834Z","signature_b64":"Vg6ShurYksBc6VtQf81PAND4kk4VHX9qzzI61XJDp9IQmD7aebAkLSHWJCcMAoH7pPFBTqB2M3Gs/y+CVp7xCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1f957d31db463260f4f66e65d7899db6e9ae18ae3c4fcff06c667dd561db483c","last_reissued_at":"2026-07-05T12:00:14.095389Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:00:14.095389Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Taming the Chaos: Coordinated Autoscaling for Heterogeneous and Disaggregated LLM Inference","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.DC","authors_text":"Chunlei Han, Huanle Han, Long Huang, Rongzhi Li, Ruogu Du, Shufan Liu, Sida Zhao, Yiwen Shao, Zefang Chu, Zherui Liu, Zuocheng Shi","submitted_at":"2025-08-27T04:22:02Z","abstract_excerpt":"Serving Large Language Models (LLMs) is a GPU-intensive task where traditional autoscalers fall short, particularly for modern Prefill-Decode (P/D) disaggregated architectures. This architectural shift, while powerful, introduces significant operational challenges, including inefficient use of heterogeneous hardware, network bottlenecks, and critical imbalances between prefill and decode stages. We introduce HeteroScale, a coordinated autoscaling framework that addresses the core challenges of P/D disaggregated serving. HeteroScale combines a topology-aware scheduler that adapts to heterogeneo"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.19559","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/2508.19559/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":"2508.19559","created_at":"2026-07-05T12:00:14.095447+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.19559v1","created_at":"2026-07-05T12:00:14.095447+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.19559","created_at":"2026-07-05T12:00:14.095447+00:00"},{"alias_kind":"pith_short_12","alias_value":"D6KX2MO3IYZG","created_at":"2026-07-05T12:00:14.095447+00:00"},{"alias_kind":"pith_short_16","alias_value":"D6KX2MO3IYZGB5HW","created_at":"2026-07-05T12:00:14.095447+00:00"},{"alias_kind":"pith_short_8","alias_value":"D6KX2MO3","created_at":"2026-07-05T12:00:14.095447+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.19271","citing_title":"TurboServe: Serving Streaming Video Generation Efficiently and Economically","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17227","citing_title":"Cloud-native and Distributed Systems for Efficient and Scalable Large Language Models -- A Research Agenda","ref_index":67,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/D6KX2MO3IYZGB5HWNZS5PCM5W3","json":"https://pith.science/pith/D6KX2MO3IYZGB5HWNZS5PCM5W3.json","graph_json":"https://pith.science/api/pith-number/D6KX2MO3IYZGB5HWNZS5PCM5W3/graph.json","events_json":"https://pith.science/api/pith-number/D6KX2MO3IYZGB5HWNZS5PCM5W3/events.json","paper":"https://pith.science/paper/D6KX2MO3"},"agent_actions":{"view_html":"https://pith.science/pith/D6KX2MO3IYZGB5HWNZS5PCM5W3","download_json":"https://pith.science/pith/D6KX2MO3IYZGB5HWNZS5PCM5W3.json","view_paper":"https://pith.science/paper/D6KX2MO3","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.19559&json=true","fetch_graph":"https://pith.science/api/pith-number/D6KX2MO3IYZGB5HWNZS5PCM5W3/graph.json","fetch_events":"https://pith.science/api/pith-number/D6KX2MO3IYZGB5HWNZS5PCM5W3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/D6KX2MO3IYZGB5HWNZS5PCM5W3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/D6KX2MO3IYZGB5HWNZS5PCM5W3/action/storage_attestation","attest_author":"https://pith.science/pith/D6KX2MO3IYZGB5HWNZS5PCM5W3/action/author_attestation","sign_citation":"https://pith.science/pith/D6KX2MO3IYZGB5HWNZS5PCM5W3/action/citation_signature","submit_replication":"https://pith.science/pith/D6KX2MO3IYZGB5HWNZS5PCM5W3/action/replication_record"}},"created_at":"2026-07-05T12:00:14.095447+00:00","updated_at":"2026-07-05T12:00:14.095447+00:00"}