{"ID":2877946,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2508.18736","arxiv_id":"2508.18736","title":"Rethinking Caching for LLM Serving Systems: Beyond Traditional Heuristics","abstract":"Serving Large Language Models (LLMs) at scale requires meeting strict Service Level Objectives (SLOs) under severe computational and memory constraints. Nevertheless, traditional caching strategies fall short: exact-matching and prefix caches neglect query semantics, while state-of-the-art semantic caches remain confined to traditional intuitions, offering little conceptual departure. Building on this, we present SISO, a semantic caching system that redefines efficiency for LLM serving. SISO introduces centroid-based caching to maximize coverage with minimal memory, locality-aware replacement to preserve high-value entries, and dynamic thresholding to balance accuracy and latency under varying workloads. Across diverse datasets, SISO delivers up to 1.71$\\times$ higher hit ratios and consistently stronger SLO attainment compared to state-of-the-art systems.","short_abstract":"Serving Large Language Models (LLMs) at scale requires meeting strict Service Level Objectives (SLOs) under severe computational and memory constraints. Nevertheless, traditional caching strategies fall short: exact-matching and prefix caches neglect query semantics, while state-of-the-art semantic caches remain confin...","url_abs":"https://arxiv.org/abs/2508.18736","url_pdf":"https://arxiv.org/pdf/2508.18736v1","authors":"[\"Jungwoo Kim\",\"Minsang Kim\",\"Jaeheon Lee\",\"Chanwoo Moon\",\"Heejin Kim\",\"Taeho Hwang\",\"Woosuk Chung\",\"Yeseong Kim\",\"Sungjin Lee\"]","published":"2025-08-26T07:09:09Z","proceeding":"cs.DB","tasks":"[\"cs.DB\",\"cs.LG\"]","methods":"[\"Large Language Model\",\"Language Model\"]","has_code":false}