{
  "slug": "request-coalescing",
  "term": "Request coalescing (cache)",
  "aliases": [
    "Cache request coalescing"
  ],
  "definition": "Cache request coalescing combines compatible pending requests for the same object behind one in-flight fetch instead of issuing duplicate backend fetches.",
  "note": "The delayed-hit model allows one outstanding fetch per object and associates later pending requests with it. They are released when the object becomes available. This reduces duplicate fetches within that model; it does not make a pending request a resident hit. [latency-pipeline]\n\nUse a key that identifies the same permitted representation and validity requirements. The single immutable-object example has no writes or invalidations and does not establish correctness for mutable data or caller-specific responses. [hesela]",
  "related": [
    "delayed-hit",
    "cache-admission",
    "sieve-eviction"
  ],
  "citations": [
    {
      "id": "latency-pipeline",
      "title": "Keren et al. Latency-Aware Caching with Delayed Hits: From Bursty Traffic to Pipeline Architectures. NSDI 2026, sections 1.1 and 4.",
      "url": "https://www.usenix.org/system/files/nsdi26-keren.pdf",
      "accessed": "2026-10-19"
    },
    {
      "id": "hesela",
      "title": "Hesela: request-coalescing assumptions in the single-object delayed-hit example.",
      "url": "https://hesela.com/analyses/sieve-cache-eviction-traces/#delayed-hits",
      "accessed": "2026-10-20"
    }
  ],
  "status": "research",
  "advantages": [],
  "limitations": [],
  "supersedes": [],
  "supersededBy": [],
  "firstSeen": null,
  "lastSeen": null,
  "updated": "2026-10-20"
}