{
  "slug": "tinylfu",
  "term": "TinyLFU",
  "aliases": [
    "Tiny Least Frequently Used"
  ],
  "definition": "TinyLFU is a cache admission policy that compares approximate recent access frequencies of a candidate and an eviction victim.",
  "note": "Periodic aging reduces the influence of old requests. Estimates cover more than resident entries and are not an exact frequency histogram. [tinylfu]\n\nWindow TinyLFU combines a resident recency window with frequency-based admission to a main cache. The window lets new bursts accumulate accesses before the comparison. Window sizing is implementation-specific. [caffeine]",
  "related": [
    "cache-admission",
    "cache-pollution",
    "lru-replacement",
    "bloom-filter"
  ],
  "citations": [
    {
      "id": "tinylfu",
      "title": "Einziger, Friedman and Manes. TinyLFU: A Highly Efficient Cache Admission Policy. arXiv:1512.00727v2 (2015), sections 3-5.",
      "url": "https://arxiv.org/pdf/1512.00727v2",
      "accessed": "2026-10-16"
    },
    {
      "id": "caffeine",
      "title": "Caffeine project. Efficiency: Window TinyLFU and adaptive admission-window sizing.",
      "url": "https://github.com/ben-manes/caffeine/wiki/Efficiency",
      "accessed": "2026-10-16"
    }
  ],
  "provenance": [
    "The TinyLFU citation is the 2015 v2 preprint, not the later journal version. It does not establish retail SSD firmware behavior. [tinylfu]"
  ],
  "status": "research",
  "advantages": [],
  "limitations": [],
  "supersedes": [],
  "supersededBy": [],
  "firstSeen": null,
  "lastSeen": null,
  "updated": "2026-10-16"
}