{"uuid": "536d1ab0-708d-4ffe-8ca2-c2202430d6c1", "vulnerability_lookup_origin": "1a89b78e-f703-45f3-bb86-59eb712668bd", "author": "9f56dd64-161d-43a6-b9c3-555944290a09", "vulnerability": "CVE-2017-5638", "type": "seen", "source": "https://gist.github.com/yuuki/2338fa88d74b49dbc0b9f62759b9ba67", "content": "# SLI/SLO \u6559\u79d1\u66f8\n\n> [!abstract] \u6982\u8981\n> \u672c\u6587\u66f8\u306f\u3001SRE \u306b\u304a\u3051\u308b SLI/SLO/SLA \u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af\u3092\u3001\u57fa\u790e\u5b9a\u7fa9\u304b\u3089\u7d44\u7e54\u7684\u5c0e\u5165\u3001\u5fdc\u7528\u9818\u57df\u307e\u3067\u7db2\u7f85\u7684\u306b\u307e\u3068\u3081\u305f\u6559\u79d1\u66f8\u3067\u3042\u308b\u3002SRE Book\u3001SRE Workbook\u3001SREcon \u5404\u5e74\u306e\u767a\u8868\u3001\u5b66\u8853\u8ad6\u6587\u3001\u56fd\u5185\u5b9f\u8df5\u4e8b\u4f8b\u306a\u3069 49 \u4ef6\u306e\u6587\u732e\u3092\u6a2a\u65ad\u7684\u306b\u7d71\u5408\u3059\u308b\u3002\n\n---\n\n## \u7b2c 0 \u7ae0\u3000\u6b74\u53f2\u7684\u8d77\u6e90\n\nSLI/SLO \u306e\u6982\u5ff5\u306f\u30bd\u30d5\u30c8\u30a6\u30a7\u30a2\u306b\u56fa\u6709\u306e\u3082\u306e\u3067\u306f\u306a\u3044\u30021984 \u5e74\u306e\u7c73\u56fd\u9ad8\u901f\u9053\u8def\u7ba1\u7406\u3067\u306f\u3001(1) Asset Quality Index(\u8def\u9762\u306e\u30e9\u30d5\u30cd\u30b9\u306a\u3069\u6750\u6599\u54c1\u8cea\u306e\u7d76\u5bfe\u5c3a\u5ea6)\u3068 (2) Customer Happiness Index(\u5e73\u5747\u6e0b\u6ede\u6642\u9593\u3001\u5e73\u5747\u8eca\u901f\u3001\u4ea4\u901a\u5bc6\u5ea6)\u3068\u3044\u3046 2 \u7a2e\u306e\u6307\u6a19\u304c\u5b9a\u7fa9\u3055\u308c\u3001\u300cLevels of Service\u300d\u3068\u3044\u3046\u7528\u8a9e\u3067\u547c\u3070\u308c\u3066\u3044\u305f\u3002\u4e0b\u6c34\u9053\u7ba1\u7406\u3067\u3082\u540c\u69d8\u306b\u3001\u30d1\u30a4\u30d7\u3042\u305f\u308a\u306e\u3072\u3073\u5272\u308c\u6570(\u8cc7\u7523\u54c1\u8cea)\u3068\u6c34\u6d17\u306e\u7a3c\u50cd\u7387(\u9867\u5ba2\u6e80\u8db3)\u304c\u533a\u5225\u3055\u308c\u3066\u3044\u305f [1]\u3002\n\n2003 \u5e74\u3001Ben Treynor Sloss\uff08Google\uff09\u304c SRE \u30c1\u30fc\u30e0\u3092\u5275\u8a2d\u3057\u3001\u30bd\u30d5\u30c8\u30a6\u30a7\u30a2\u30a8\u30f3\u30b8\u30cb\u30a2\u30ea\u30f3\u30b0\u306e\u624b\u6cd5\u3092\u904b\u7528\u306b\u9069\u7528\u3059\u308b\u30c7\u30a3\u30b7\u30d7\u30ea\u30f3\u3092\u78ba\u7acb\u3057\u305f\u3002SLI/SLO/SLA \u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af\u3068\u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u306f\u30012014 \u5e74\u306e SREcon14 \u57fa\u8abf\u8b1b\u6f14\u3067\u521d\u3081\u3066\u516c\u306b\u4f53\u7cfb\u5316\u3055\u308c [2]\u30012016 \u5e74\u306e SRE Book \u3067\u7406\u8ad6\u7684\u57fa\u76e4\u304c\u78ba\u7acb\u3055\u308c\u305f [3]\u3002\n\n---\n\n## \u7b2c 1 \u7ae0\u3000\u57fa\u790e\u6982\u5ff5 \u2014 SLI/SLO/SLA \u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af\n\n### 1.1 SLI(\u30b5\u30fc\u30d3\u30b9\u30ec\u30d9\u30eb\u6307\u6a19)\n\nSLI \u306f\u30b5\u30fc\u30d3\u30b9\u306e\u632f\u308b\u821e\u3044\u3092\u5b9a\u91cf\u7684\u306b\u8a08\u6e2c\u3059\u308b\u6307\u6a19\u3067\u3042\u308b\u3002\u30ea\u30af\u30a8\u30b9\u30c8\u30ec\u30a4\u30c6\u30f3\u30b7\u3001\u30a8\u30e9\u30fc\u7387\u3001\u30b9\u30eb\u30fc\u30d7\u30c3\u30c8\u3001\u53ef\u7528\u6027\u306a\u3069\u304c\u5178\u578b\u4f8b\u306b\u3042\u305f\u308b\u3002SLI \u306f\u30b5\u30fc\u30d3\u30b9\u306e\u5065\u5168\u6027\u3092\u76f4\u63a5\u53cd\u6620\u3059\u308b**\u8a08\u6e2c\u5024**\u3067\u3042\u308a\u3001\u76ee\u6a19\u5024\u3067\u306f\u306a\u3044 [3]\u3002\n\nSRE Workbook \u306f SLI \u3092\u300c\u826f\u3044\u30a4\u30d9\u30f3\u30c8\u6570 / \u5168\u30a4\u30d9\u30f3\u30c8\u6570\u300d\u306e\u6bd4\u7387\u3068\u3057\u3066\u8a2d\u8a08\u3059\u308b\u3053\u3068\u3092\u63a8\u5968\u3059\u308b\u30020% \u306f\u4f55\u3082\u52d5\u304b\u306a\u3044\u72b6\u614b\u3001100% \u306f\u4f55\u3082\u58ca\u308c\u3066\u3044\u306a\u3044\u72b6\u614b\u3068\u3044\u3046\u76f4\u611f\u7684\u306a\u5c3a\u5ea6\u306b\u306a\u308a\u3001SLO \u3068\u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u306b\u63a5\u7d9a\u3057\u3084\u3059\u3044 [4]\u3002\n\n\u91cd\u8981\u306a\u8a2d\u8a08\u539f\u5247\u3068\u3057\u3066\u3001SLI \u306f**\u4ed5\u69d8**\u3068**\u5b9f\u88c5**\u306b\u5206\u3051\u3066\u8003\u3048\u308b\u3002\n\n- **SLI \u4ed5\u69d8**: \u30e6\u30fc\u30b6\u30fc\u306b\u3068\u3063\u3066\u610f\u5473\u306e\u3042\u308b\u8a55\u4fa1(\u4f8b:\u300c\u30ea\u30af\u30a8\u30b9\u30c8\u304c 200ms \u4ee5\u5185\u306b\u6b63\u5e38\u5fdc\u7b54\u3092\u8fd4\u3059\u300d)\n- **SLI \u5b9f\u88c5**: \u305d\u308c\u3092\u3069\u3053\u3067\u3069\u3046\u6e2c\u308b\u304b(\u4f8b:\u300c\u30ed\u30fc\u30c9\u30d0\u30e9\u30f3\u30b5\u306e\u30a2\u30af\u30bb\u30b9\u30ed\u30b0\u3067 HTTP 5xx \u4ee5\u5916\u3092\u6210\u529f\u3001\u30ec\u30b9\u30dd\u30f3\u30b9\u30bf\u30a4\u30e0 200ms \u4ee5\u4e0b\u3092\u9ad8\u901f\u3068\u6570\u3048\u308b\u300d)\n\n### 1.2 SLO(\u30b5\u30fc\u30d3\u30b9\u30ec\u30d9\u30eb\u76ee\u6a19)\n\nSLO \u306f SLI \u306e\u76ee\u6a19\u5024\u3042\u308b\u3044\u306f\u8a31\u5bb9\u7bc4\u56f2\u3067\u3042\u308b\u3002\u300cSLI \u2264 target\u300d\u306e\u5f62\u3092\u3068\u308b\u3002\u4f8b:\u300c\u30ea\u30af\u30a8\u30b9\u30c8\u30ec\u30a4\u30c6\u30f3\u30b7\u306e p99 \u304c 200ms \u4ee5\u4e0b\u300d\u3002SLO \u306f\u30b5\u30fc\u30d3\u30b9\u306e\u4fe1\u983c\u6027\u306b\u95a2\u3059\u308b**\u5185\u90e8\u7684\u306a\u76ee\u6a19**\u3067\u3042\u308a\u3001\u5916\u90e8\u3078\u306e\u7d04\u675f\u3067\u306f\u306a\u3044 [3]\u3002\n\n### 1.3 SLA(\u30b5\u30fc\u30d3\u30b9\u30ec\u30d9\u30eb\u5408\u610f)\n\nSLA \u306f SLO \u306b**\u9055\u53cd\u6642\u306e\u7f70\u5247(consequences)**\u3092\u4ed8\u52a0\u3057\u305f\u660e\u793a\u7684\u306a\u5951\u7d04\u3067\u3042\u308b\u3002SLO + \u5e30\u7d50 = SLA \u3068\u5b9a\u5f0f\u5316\u3055\u308c\u308b [5]\u3002SLA \u306f\u901a\u5e38\u3001\u5916\u90e8\u9867\u5ba2\u3068\u306e\u9593\u3067\u7de0\u7d50\u3055\u308c\u308b\u3002\n\nSLI \u2192 SLO \u2192 SLA \u306f\u62bd\u8c61\u5ea6\u304c\u4e0a\u304c\u308b\u65b9\u5411\u306e\u968e\u5c64\u3092\u306a\u3059\u3002SLI \u306a\u3057\u306b SLO \u306f\u5b9a\u7fa9\u3067\u304d\u305a\u3001SLO \u306a\u3057\u306b SLA \u306f\u610f\u5473\u3092\u6301\u305f\u306a\u3044\u3002\n\n### 1.4 \u306a\u305c 100% \u3092\u76ee\u6307\u3055\u306a\u3044\u306e\u304b\n\nSRE \u306e\u6838\u5fc3\u539f\u5247\u306e\u4e00\u3064\u306f\u3001100% \u306e\u53ef\u7528\u6027\u3092\u8ffd\u6c42\u3057\u306a\u3044\u3053\u3068\u3067\u3042\u308b\u3002\u7406\u7531\u306f 3 \u3064\u3042\u308b\u3002\n\n1. \u30e6\u30fc\u30b6\u30fc\u306f 99.99% \u3068 99.999% \u306e\u5dee\u3092\u691c\u77e5\u3067\u304d\u306a\u3044\n2. \u4fe1\u983c\u6027\u6539\u5584\u306e\u30b3\u30b9\u30c8\u66f2\u7dda\u306f\u975e\u7dda\u5f62\u3067\u3042\u308a\u3001\u5404\u5897\u5206\u304c\u524d\u56de\u306e 100 \u500d\u306e\u30b3\u30b9\u30c8\u3092\u8981\u3059\u308b\n3. \u9867\u5ba2\u5074\u306e\u4f9d\u5b58\u95a2\u4fc2(\u7aef\u672b\u3001\u30cd\u30c3\u30c8\u30ef\u30fc\u30af)\u3082\u542b\u3081\u308b\u3068\u3001100% \u306e\u4f53\u9a13\u306f\u7269\u7406\u7684\u306b\u4e0d\u53ef\u80fd\n\n[6]\n\n---\n\n## \u7b2c 2 \u7ae0\u3000SLI \u306e\u8a2d\u8a08 \u2014 \u4f55\u3092\u30fb\u3069\u3053\u3067\u30fb\u3069\u3046\u6e2c\u308b\u304b\n\n### 2.1 \u30b5\u30fc\u30d3\u30b9\u7a2e\u5225\u3054\u3068\u306e\u512a\u5148\u6307\u6a19\n\n| \u30b5\u30fc\u30d3\u30b9\u7a2e\u5225 | \u512a\u5148 SLI |\n|---|---|\n| \u30e6\u30fc\u30b6\u30fc\u5411\u3051\u30b5\u30fc\u30d3\u30b9 | \u53ef\u7528\u6027\u3001\u30ec\u30a4\u30c6\u30f3\u30b7\u3001\u30b9\u30eb\u30fc\u30d7\u30c3\u30c8 |\n| \u30b9\u30c8\u30ec\u30fc\u30b8\u30b7\u30b9\u30c6\u30e0 | \u8010\u4e45\u6027\u3001\u53ef\u7528\u6027\u3001\u30ec\u30a4\u30c6\u30f3\u30b7 |\n| \u30d3\u30c3\u30b0\u30c7\u30fc\u30bf\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3 | \u30b9\u30eb\u30fc\u30d7\u30c3\u30c8\u3001\u30a8\u30f3\u30c9\u30c4\u30fc\u30a8\u30f3\u30c9\u30ec\u30a4\u30c6\u30f3\u30b7 |\n\n[3]\n\nSRE Workbook \u306f\u3001\u30ea\u30af\u30a8\u30b9\u30c8\u99c6\u52d5\u30b5\u30fc\u30d3\u30b9\u306b\u306f\u6210\u529f\u5fdc\u7b54\u7387\u30fb\u95be\u5024\u4ee5\u4e0b\u30ec\u30a4\u30c6\u30f3\u30b7\u30fb\u52a3\u5316\u306a\u3057\u5fdc\u7b54\u306e\u5272\u5408\u3092\u3001\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3\u306b\u306f\u30c7\u30fc\u30bf\u9bae\u5ea6\u30fb\u6b63\u78ba\u6027\u30fb\u30ab\u30d0\u30ec\u30c3\u30b8\u3092\u3001\u30b9\u30c8\u30ec\u30fc\u30b8\u306b\u306f\u66f8\u304d\u8fbc\u3093\u3060\u30ec\u30b3\u30fc\u30c9\u304c\u8aad\u307f\u51fa\u305b\u308b\u5272\u5408\u3092\u63a8\u5968\u3059\u308b [4]\u3002\n\n### 2.2 \u30d1\u30fc\u30bb\u30f3\u30bf\u30a4\u30eb\u306e\u91cd\u8996\n\n\u30ec\u30a4\u30c6\u30f3\u30b7\u6307\u6a19\u3092**\u5e73\u5747\u5024**\u3067\u8a55\u4fa1\u3057\u3066\u306f\u306a\u3089\u306a\u3044\u3002\u5e73\u5747\u306f\u30c6\u30fc\u30eb\u30ec\u30a4\u30c6\u30f3\u30b7\u306e\u60c5\u5831\u3092\u6f70\u3059\u3002\u63a8\u5968\u3055\u308c\u308b\u8a08\u6e2c\u306f\u30d1\u30fc\u30bb\u30f3\u30bf\u30a4\u30eb\u3067\u3042\u308a\u3001\u7279\u306b p50(\u5178\u578b\u7684\u4f53\u9a13)\u3001p99(\u30c6\u30fc\u30eb)\u3001p99.9(\u6700\u60aa\u4f53\u9a13)\u304c\u91cd\u8996\u3055\u308c\u308b [3]\u3002\n\n\u305f\u3060\u3057\u30d1\u30fc\u30bb\u30f3\u30bf\u30a4\u30eb\u306b\u306f\u91cd\u5927\u306a\u5b9f\u88c5\u4e0a\u306e\u5236\u7d04\u304c\u3042\u308b\u3002Hartmann\uff08Circonus\uff09\u306f\u3001\u6642\u9593\u3054\u3068\u306e p90 \u3092\u5e73\u5747\u3059\u308b\u3068\u5168\u4f53\u5206\u5e03\u304b\u3089\u8a08\u7b97\u3057\u305f p90 \u3068\u305a\u308c\u308b\u3053\u3068\u3092\u793a\u3057\u3001**\u30d1\u30fc\u30bb\u30f3\u30bf\u30a4\u30eb\u306f\u8907\u6570\u6642\u9593\u7a93\u30fb\u8907\u6570\u30ce\u30fc\u30c9\u3092\u307e\u305f\u3044\u3067\u96c6\u7d04\u3067\u304d\u306a\u3044**\u3053\u3068\u3092\u5b9f\u52d9\u4e0a\u306e\u554f\u984c\u3068\u3057\u3066\u5b9a\u5f0f\u5316\u3057\u305f\u3002\u89e3\u6c7a\u7b56\u306f\u300c\u3057\u304d\u3044\u5024\u4ee5\u5185\u306e\u30ea\u30af\u30a8\u30b9\u30c8\u6570 / \u5168\u30ea\u30af\u30a8\u30b9\u30c8\u6570\u300d\u3068\u3044\u3046\u826f\u3044\u30a4\u30d9\u30f3\u30c8\u6bd4\u7387\u306b\u5909\u63db\u3059\u308b\u3053\u3068\u3067\u3042\u308b [7]\u3002\n\n### 2.3 \u6e2c\u5b9a\u70b9\u306e\u9078\u629e\n\nJones\u30fbMurphy\uff08Google\uff09\u306f SREcon16 \u3067\u3001\u30b5\u30fc\u30d0\u30fc\u5074\u30ec\u30a4\u30c6\u30f3\u30b7\u3092\u6e2c\u308a\u304c\u3061\u306a Google \u306e\u5b9f\u614b\u3092\u300c\u9593\u9055\u3044\u300d\u3068\u8a8d\u3081\u3001\u30af\u30e9\u30a4\u30a2\u30f3\u30c8\u5074\u304c\u5b9f\u969b\u306e\u30e6\u30fc\u30b6\u30fc\u4f53\u9a13\u3092\u6b63\u78ba\u306b\u53cd\u6620\u3059\u308b\u3068\u8ff0\u3079\u305f\u3002**\u5236\u5fa1\u3057\u3084\u3059\u3044\u5834\u6240\u3067\u6e2c\u308b(locus of control)\u306e\u3068\u3001\u30e6\u30fc\u30b6\u30fc\u306b\u610f\u5473\u306e\u3042\u308b\u5834\u6240\u3067\u6e2c\u308b(locus of measurement)\u306e\u306f\u7570\u306a\u308b** [8]\u3002\n\nSLI \u5b9f\u88c5\u306e\u9078\u629e\u80a2\u306b\u306f\u3001\u30b5\u30fc\u30d0\u30fc\u30ed\u30b0\u30fb\u30ed\u30fc\u30c9\u30d0\u30e9\u30f3\u30b5\u76e3\u8996\u30fb\u30d6\u30e9\u30c3\u30af\u30dc\u30c3\u30af\u30b9\u76e3\u8996\u30fb\u30af\u30e9\u30a4\u30a2\u30f3\u30c8\u5074\u8a08\u88c5\u304c\u3042\u308a\u3001\u305d\u308c\u305e\u308c\u54c1\u8cea\u30fb\u30ab\u30d0\u30ec\u30c3\u30b8\u30fb\u30b3\u30b9\u30c8\u306e\u30c8\u30ec\u30fc\u30c9\u30aa\u30d5\u304c\u3042\u308b [4]\u3002\n\n### 2.4 \u30b1\u30a4\u30d1\u30d3\u30ea\u30c6\u30a3\u99c6\u52d5\u306e SLI \u8a2d\u8a08\n\nNew Relic \u306e Binette\u30fbFlaming \u306f\u3001\u30b7\u30b9\u30c6\u30e0\u5883\u754c\u3054\u3068\u306b\u300c\u516c\u958b\u3059\u308b\u30b1\u30a4\u30d1\u30d3\u30ea\u30c6\u30a3\u300d\u3092\u5217\u6319\u3057\u3001\u5404\u30b1\u30a4\u30d1\u30d3\u30ea\u30c6\u30a3\u306e\u300c\u5229\u7528\u53ef\u80fd\u300d\u3092\u5e73\u6613\u306a\u8a00\u8449\u3067\u5b9a\u7fa9\u3057\u3066\u304b\u3089\u6280\u8853\u7684 SLI \u306b\u843d\u3068\u3059 7 \u30b9\u30c6\u30c3\u30d7\u30ec\u30b7\u30d4\u3092\u793a\u3057\u305f\u3002SLI \u8a2d\u8a08\u3067\u300c\u4f55\u3092\u6e2c\u308c\u3070\u3044\u3044\u304b\u308f\u304b\u3089\u306a\u3044\u300d\u3068\u3044\u3046\u96e3\u70b9\u3092\u3001**\u300c\u30b7\u30b9\u30c6\u30e0\u5883\u754c\u306e\u6a5f\u80fd\u4e00\u89a7\u3092\u4f5c\u308b\u300d\u3068\u3044\u3046\u64cd\u4f5c\u53ef\u80fd\u306a\u7b2c\u4e00\u6b69\u306b\u5909\u63db**\u3059\u308b [9]\u3002\n\n### 2.5 \u300c\u6210\u529f\u300d\u3092\u5b9a\u7fa9\u3059\u3079\u304d\u7406\u7531\n\nSLI \u306f\u300c\u60aa\u3044\u6319\u52d5\u300d\u3067\u306f\u306a\u304f\u300c\u6210\u529f\u300d\u3092\u5b9a\u7fa9\u3059\u3079\u304d\u3067\u3042\u308b\u3002\u30a8\u30e9\u30fc\u3092\u5b9a\u7fa9\u3059\u308b\u5834\u5408\u306f\u65e2\u77e5\u306e\u554f\u984c\u306e\u307f\u3092\u30ab\u30d0\u30fc\u3059\u308b(\u6709\u9650\u30b9\u30b3\u30fc\u30d7)\u304c\u3001\u6210\u529f\u3092\u5b9a\u7fa9\u3059\u308c\u3070\u300c\u305d\u308c\u4ee5\u5916\u306f\u3059\u3079\u3066\u4e0d\u78ba\u304b\u300d\u3068\u3044\u3046\u5305\u62ec\u7684\u5b9a\u7fa9\u306b\u306a\u308b [10]\u3002\n\n### 2.6 \u30e6\u30fc\u30b6\u30fc\u5e78\u798f\u306e 6 \u30d5\u30ec\u30fc\u30d0\u30fc\n\nGangatirkar\uff08Indeed\uff09\u306f SLI \u5019\u88dc\u3092 6 \u7a2e\u306b\u6574\u7406\u3057\u305f: Availability / Responsiveness / Freshness / Completeness / Accuracy / Breadth\u3002\u3053\u306e\u67a0\u7d44\u307f\u3068\u30b1\u30a4\u30d1\u30d3\u30ea\u30c6\u30a3\u5217\u6319\u3092\u7d44\u307f\u5408\u308f\u305b\u308b\u3068\u3001SLI \u5019\u88dc\u306e\u898b\u843d\u3068\u3057\u3092\u9632\u304e\u3064\u3064\u30b5\u30fc\u30d3\u30b9\u56fa\u6709\u306b\u5177\u4f53\u5316\u3067\u304d\u308b [11]\u3002\n\n### 2.7 \u96c6\u7d04\u30a4\u30f3\u30bf\u30fc\u30d0\u30eb\u306e\u554f\u984c\n\nSREcon16 \u3067 Jones \u306f\u3001\u540c\u3058\u30c7\u30fc\u30bf\u304c 1 \u5206\u30a4\u30f3\u30bf\u30fc\u30d0\u30eb\u3067\u306f\u5e73\u5766\u300130 \u79d2\u3067\u306f\u6563\u767a\u7684\u30b9\u30d1\u30a4\u30af\u30011 \u79d2\u3067\u306f\u300c\u6bce\u5206\u5148\u982d\u306b\u5168\u30ea\u30af\u30a8\u30b9\u30c8\u304c\u96c6\u4e2d\u300d\u3068\u898b\u3048\u308b\u3053\u3068\u3092\u793a\u3057\u305f\u3002SLI \u8a2d\u8a08\u306b\u304a\u3044\u3066\u300c\u4f55\u3092\u6e2c\u308b\u304b\u300d\u306e\u524d\u306b\u300c**\u3069\u306e\u30a6\u30a3\u30f3\u30c9\u30a6\u3067\u898b\u308b\u304b**\u300d\u3068\u3044\u3046\u554f\u3044\u304c\u3042\u308b [8]\u3002\n\n---\n\n## \u7b2c 3 \u7ae0\u3000SLO \u306e\u8a2d\u5b9a \u2014 \u76ee\u6a19\u5024\u306e\u6c7a\u3081\u65b9\n\n### 3.1 \u521d\u671f SLO \u306f\u6b63\u89e3\u3067\u306a\u304f\u3066\u3088\u3044\n\n\u521d\u671f SLO \u306f\u6b63\u3057\u304f\u306a\u304f\u3066\u3088\u3044\u3002\u5b89\u4fa1\u306b\u6e2c\u308c\u308b\u6307\u6a19\u304b\u3089\u59cb\u3081\u3001\u30e6\u30fc\u30b6\u30fc\u6e80\u8db3\u30fb\u65e2\u77e5\u969c\u5bb3\u30fb\u30b5\u30dd\u30fc\u30c8\u30c1\u30b1\u30c3\u30c8\u30fb\u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u640d\u5931\u3068\u306e\u76f8\u95a2\u3067\u7d99\u7d9a\u6539\u5584\u3059\u308b [4]\u3002\n\nEvernote \u306f\u6708\u6b21 99.95% \u7a3c\u50cd\u7387\u3068\u3044\u3046\u5358\u7d14\u306a\u521d\u671f SLO \u3067\u958b\u59cb\u3057\u3001\u6708\u6b21\u30ec\u30d3\u30e5\u30fc\u3068\u534a\u5e74\u3054\u3068\u306e\u898b\u76f4\u3057\u3067\u6539\u5584\u3057\u305f [12]\u3002\n\n### 3.2 S \u5b57\u66f2\u7dda\u3067\u75db\u307f\u306e\u3057\u304d\u3044\u5024\u3092\u7279\u5b9a\u3059\u308b\n\nGangatirkar \u306f\u53ef\u7528\u6027(\u6a2a\u8ef8)\u5bfe\u30e6\u30fc\u30b6\u30fc\u5e78\u798f\u5ea6(\u7e26\u8ef8)\u306e S \u5b57\u66f2\u7dda\u3092\u7528\u3044\u3001SLO \u306f\u30e6\u30fc\u30b6\u30fc\u304c\u4e0d\u6e80\u3092\u611f\u3058\u59cb\u3081\u308b\u524d\u306e\u6c34\u6e96\u306b\u8a2d\u5b9a\u3059\u3079\u304d\u3060\u3068\u3057\u305f\u3002\u5c48\u66f2\u70b9\u304c\u75db\u307f\u306e\u3057\u304d\u3044\u5024\u3067\u3042\u308a\u3001SLO \u306f\u305d\u308c\u3088\u308a\u4e0b\u306b\u8a2d\u5b9a\u3059\u308b [11]\u3002\n\n### 3.3 \u5171\u611f\u30ae\u30e3\u30c3\u30d7\n\nGangatirkar \u306e 2\u00d72 \u30de\u30c8\u30ea\u30af\u30b9\u306f\u91cd\u8981\u306a\u6982\u5ff5\u3092\u5b9a\u7fa9\u3059\u308b\u3002\n\n| | SLO \u9054\u6210 | SLO \u672a\u9054 |\n|---|---|---|\n| **\u30e6\u30fc\u30b6\u30fc\u6e80\u8db3** | \u6b63\u5e38 | \u904e\u5270\u9054\u6210\u306e\u5e7b\u5f71 |\n| **\u30e6\u30fc\u30b6\u30fc\u4e0d\u6e80** | **\u5171\u611f\u30ae\u30e3\u30c3\u30d7** | \u6b63\u5e38 |\n\n\u300cSLO \u9054\u6210\u30fb\u30e6\u30fc\u30b6\u30fc\u4e0d\u6e80\u300d\u306e\u30bb\u30eb\u304c\u5171\u611f\u30ae\u30e3\u30c3\u30d7(Empathy Gap)\u3067\u3042\u308a\u3001SLO \u304c\u30e6\u30fc\u30b6\u30fc\u306e\u75db\u307f\u306e\u3057\u304d\u3044\u5024\u3088\u308a\u7de9\u304f\u8a2d\u5b9a\u3055\u308c\u3066\u3044\u308b\u5834\u5408\u306b\u767a\u751f\u3059\u308b [11]\u3002\n\n### 3.4 SLO \u306e Rationale \u3092\u30e6\u30fc\u30b6\u30fc\u884c\u52d5\u30c7\u30fc\u30bf\u3067\u6839\u62e0\u3065\u3051\u308b\n\nStanke\uff08Google\uff09\u306f SLO Policy \u306b \"Error rates greater than .05% correlate with significant increase in customer support tickets\" \u3068\u3044\u3046 Rationale \u3092\u6dfb\u3048\u3001\u6280\u8853\u95be\u5024\u304c\u9867\u5ba2\u884c\u52d5\u6307\u6a19\u3068\u7d50\u3073\u3064\u3044\u305f\u6839\u62e0\u3092\u6301\u3064\u3053\u3068\u3092\u793a\u3057\u305f [13]\u3002\n\n### 3.5 \u591a\u6bb5\u30d1\u30fc\u30bb\u30f3\u30bf\u30a4\u30eb\u5b9a\u7fa9\n\n\u5358\u4e00\u306e\u300c99% \u306e\u5fdc\u7b54\u304c X \u79d2\u4ee5\u5185\u300d\u3067\u306f\u3001X \u79d2\u3092\u8d85\u3048\u308b 1% \u306e\u5206\u5e03\u3092\u5236\u5fa1\u3067\u304d\u306a\u3044\u3002\u591a\u6bb5\u5b9a\u7fa9(90%/30s\u30fb99%/1min\u30fb99.9%/5min)\u306f\u300c\u901f\u3044\u5fdc\u7b54\u306e\u5927\u591a\u6570\u300d\u3068\u300c\u9045\u3044\u5fdc\u7b54\u306e\u5c3e\u90e8\u300d\u3092\u5225\u3005\u306b\u5236\u5fa1\u3059\u308b\u3002SLO \u3092\u300c\u3057\u304d\u3044\u5024\u306e\u901a\u904e/\u672a\u901a\u904e\u300d\u304b\u3089\u300c\u5206\u5e03\u306e\u5236\u5fa1\u300d\u3068\u3057\u3066\u6271\u3046\u30a2\u30d7\u30ed\u30fc\u30c1\u3067\u3042\u308b [14][15]\u3002\n\n### 3.6 SLO \u516c\u8868\u306e\u91cd\u8981\u6027\n\nJones\u30fbMurphy \u306f SREcon16 \u3067\u3001**SLO \u3092\u516c\u958b\u3057\u306a\u3044\u30b5\u30fc\u30d3\u30b9\u3067\u306f\u30e6\u30fc\u30b6\u30fc\u304c\u300c\u4eca\u306e\u6027\u80fd\u304c\u6c38\u9060\u306b\u7d9a\u304f\u300d\u3068\u671f\u5f85\u3057\u59cb\u3081\u3001\u5c06\u6765\u306e\u8a2d\u8a08\u5909\u66f4\u30fb\u30b3\u30b9\u30c8\u6700\u9069\u5316\u306e\u9078\u629e\u80a2\u304c\u5931\u308f\u308c\u308b**\u3068\u8b66\u544a\u3057\u305f\u3002\u5bfe\u51e6\u3068\u3057\u3066\u300c\u610f\u56f3\u7684\u306a\u30c0\u30a6\u30f3\u30bf\u30a4\u30e0\u3092\u5165\u308c\u308b\u30b7\u30b9\u30c6\u30e0\u304c Google \u306b\u5b9f\u5728\u3059\u308b\u300d\u3068\u660e\u304b\u3057\u305f [8]\u3002\n\n### 3.7 SLO \u6587\u66f8\u306e\u69cb\u6210\n\nAppendix A \u306e Example Game Service \u306f\u3001SLO \u6587\u66f8\u306e\u69cb\u6210\u8981\u7d20\u3092\u793a\u3059\u3002\n\n- \u30b5\u30fc\u30d3\u30b9\u6982\u8981\u3068\u5bfe\u8c61\u30ab\u30c6\u30b4\u30ea\n- SLI \u5b9a\u7fa9(\u4f55\u3092\u3001\u3069\u3053\u3067\u6e2c\u308b\u304b)\n- SLO \u76ee\u6a19\u5024\u3068\u8a55\u4fa1\u30a6\u30a3\u30f3\u30c9\u30a6(4 \u9031\u9593\u30ed\u30fc\u30ea\u30f3\u30b0)\n- \u6839\u62e0(\u5c65\u6b74\u30c7\u30fc\u30bf\u3068\u4e38\u3081\u30eb\u30fc\u30eb)\n- \u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u8a08\u7b97\n- \u4f46\u3057\u66f8\u304d\u3068\u9650\u754c\n\nSLO \u306f\u53ef\u7528\u6027\u3060\u3051\u3067\u306a\u304f\u3001\u30ec\u30a4\u30c6\u30f3\u30b7\u3001\u9bae\u5ea6\u3001\u6b63\u78ba\u6027\u3001\u5b8c\u5168\u6027\u306a\u3069\u3001\u30e6\u30fc\u30b6\u30fc\u4fa1\u5024\u306b\u5bfe\u5fdc\u3059\u308b\u8907\u6570\u6b21\u5143\u3067\u5b9a\u7fa9\u3067\u304d\u308b [16]\u3002\n\n---\n\n## \u7b2c 4 \u7ae0\u3000\u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8 \u2014 SLO \u3092\u7d44\u7e54\u7684\u610f\u601d\u6c7a\u5b9a\u3078\u63a5\u7d9a\u3059\u308b\n\n### 4.1 \u5b9a\u7fa9\u3068\u539f\u7406\n\n\u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u3068\u306f\u3001SLO \u3067\u8a31\u5bb9\u3055\u308c\u308b\u969c\u5bb3\u91cf\u306e\u4e0a\u9650\u3092\u300c\u4e88\u7b97\u300d\u3068\u3057\u3066\u6271\u3044\u3001\u958b\u767a\u30c1\u30fc\u30e0\u3068 SRE \u304c\u5171\u6709\u3059\u308b\u4fe1\u983c\u6027\u7ba1\u7406\u306e\u4ed5\u7d44\u307f\u3067\u3042\u308b\u3002SLO \u304c 99.9% \u306a\u3089\u3070 0.1% \u306e\u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u3092\u6301\u3064\u3002\u4e88\u7b97\u3092\u6d88\u8cbb\u3057\u304d\u3063\u305f\u5834\u5408\u306f\u30ea\u30ea\u30fc\u30b9\u3092\u51cd\u7d50\u3057\u3066\u4fe1\u983c\u6027\u6539\u5584\u306b\u96c6\u4e2d\u3059\u308b [6]\u3002\n\n\u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u306e\u6700\u3082\u91cd\u8981\u306a\u7d44\u7e54\u52b9\u679c\u306f\u3001SRE \u304c\u300c\u30ce\u30fc\u300d\u304b\u3089\u300c\u30a4\u30a8\u30b9\u3001\u3082\u3057\u301c\u306a\u3089\u3070\u300d\u3078\u8ee2\u63db\u3059\u308b\u3053\u3068\u3067\u3042\u308b\u3002\u30d7\u30ed\u30c0\u30af\u30c8\u958b\u767a\u30c1\u30fc\u30e0\u304c SRE \u3092\u8fc2\u56de\u3057\u3088\u3046\u3068\u3057\u306a\u304f\u306a\u308b\u306e\u306f\u3001\u8fc2\u56de\u3059\u308b\u3068\u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u3092\u81ea\u3089\u6d88\u8cbb\u3059\u308b\u3060\u3051\u3060\u304b\u3089\u3067\u3042\u308b [8]\u3002\n\n### 4.2 \u8d77\u6e90\n\n\u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u306e\u8d77\u6e90\u306f\u300cSLA \u3092\u8d85\u3048\u3059\u304e\u308b\u3053\u3068\u3078\u306e\u6c17\u3065\u304d\u300d\u3067\u3042\u308b\u3002Alvidrez\uff08Google\uff09\u306f SREcon15 \u3067\u3001AdSense \u304c 2009 \u5e74\u306b SLA \u3092\u4e00\u8cab\u3057\u3066\u8d85\u3048\u305f\u72b6\u614b\u3092\u300c\u6a5f\u4f1a\u306e\u640d\u5931\u300d\u3068\u8a9e\u3063\u305f\u3002\u8d85\u904e\u5206\u306f\u300c\u901f\u304f\u52d5\u304f\u30fb\u5c11\u306a\u3044\u30ea\u30bd\u30fc\u30b9\u3067\u52d5\u304f\u300d\u305f\u3081\u306b\u4f7f\u3048\u305f\u3002\u3053\u306e\u9006\u8ee2\u306e\u767a\u60f3\u304c\u6982\u5ff5\u306e\u500b\u4eba\u7684\u306a\u8d77\u6e90\u3067\u3042\u308b [17]\u3002\n\n### 4.3 \u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u30dd\u30ea\u30b7\u30fc\n\nSRE Workbook Appendix B \u306f\u5177\u4f53\u7684\u306a\u30dd\u30ea\u30b7\u30fc\u4f8b\u3092\u793a\u3059\u3002\n\n- \u76f4\u8fd1 4 \u9031\u9593\u3067\u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u3092\u8d85\u904e \u2192 P0\u30fb\u30bb\u30ad\u30e5\u30ea\u30c6\u30a3\u4fee\u6b63\u4ee5\u5916\u306e\u5909\u66f4\u505c\u6b62\n- \u5358\u4e00\u30a4\u30f3\u30b7\u30c7\u30f3\u30c8\u304c 4 \u9031\u9593\u4e88\u7b97\u306e 20% \u8d85\u6d88\u8cbb \u2192 \u30dd\u30b9\u30c8\u30e2\u30fc\u30c6\u30e0\u3068 P0 \u30a2\u30af\u30b7\u30e7\u30f3\u30a2\u30a4\u30c6\u30e0\n- \u5358\u4e00\u969c\u5bb3\u30af\u30e9\u30b9\u304c\u56db\u534a\u671f\u4e88\u7b97\u306e 20% \u8d85\u6d88\u8cbb \u2192 \u7fcc\u56db\u534a\u671f\u8a08\u753b\u306b P0 \u9805\u76ee\n\n\u91cd\u8981\u306a\u89e3\u91c8: \u3053\u306e\u65b9\u91dd\u306f\u7f70\u3067\u306f\u306a\u3044\u3002\u5909\u66f4\u505c\u6b62\u306f\u671b\u307e\u3057\u3044\u72b6\u614b\u3067\u306f\u306a\u3044\u304c\u3001\u4fe1\u983c\u6027\u30c7\u30fc\u30bf\u304c\u300c\u4eca\u306f\u6a5f\u80fd\u3088\u308a\u4fe1\u983c\u6027\u3092\u512a\u5148\u3059\u3079\u304d\u300d\u3068\u793a\u3059\u3068\u304d\u3001\u30c1\u30fc\u30e0\u304c\u4fe1\u983c\u6027\u306b\u96c6\u4e2d\u3059\u308b\u305f\u3081\u306e**\u5236\u5ea6\u7684\u8a31\u53ef**\u3068\u3057\u3066\u6a5f\u80fd\u3059\u308b [18]\u3002\n\n### 4.4 \u30d0\u30f3\u30d0\u30f3\u5236\u5fa1\u304b\u3089\u30d7\u30ed\u30dd\u30fc\u30b7\u30e7\u30ca\u30eb\u5236\u5fa1\u3078\n\nJones \u306f SREcon16 \u3067\u3001\u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u3092\u4e8c\u5024\u5207\u308a\u66ff\u3048(\u300c\u30ea\u30ea\u30fc\u30b9\u5168\u958b\u300d\u2192\u300c\u30ea\u30ea\u30fc\u30b9\u5168\u505c\u6b62\u300d)\u3067\u306f\u306a\u304f\u3001\u30d0\u30fc\u30f3\u7387\u306e\u9023\u7d9a\u7684\u306a\u76e3\u8996\u306b\u57fa\u3065\u304f\u901f\u5ea6\u5236\u5fa1\u3068\u3057\u3066\u4f7f\u3046\u3079\u304d\u3060\u3068\u8aac\u660e\u3057\u305f\u3002\u5236\u5fa1\u5de5\u5b66\u3067\u3044\u3046\u300c\u30d0\u30f3\u30d0\u30f3\u5236\u5fa1\u300d\u304b\u3089\u300c\u30d7\u30ed\u30dd\u30fc\u30b7\u30e7\u30ca\u30eb\u5236\u5fa1\u300d\u3078\u306e\u79fb\u884c\u3067\u3042\u308b [8]\u3002\n\n### 4.5 \u6642\u9593\u30b9\u30e9\u30a4\u30b9 vs \u30a4\u30d9\u30f3\u30c8\u30d9\u30fc\u30b9\u96c6\u8a08\n\n99.95 SLO + 95% \u95be\u5024\u306b\u3088\u308b\u6642\u9593\u30b9\u30e9\u30a4\u30b9\u65b9\u5f0f\u3067\u306f\u3001\u30d4\u30fc\u30af\u6642(1 \u5206 1,000 \u30ea\u30af\u30a8\u30b9\u30c8)\u3068\u6df1\u591c(1 \u5206 10 \u30ea\u30af\u30a8\u30b9\u30c8)\u3092\u540c\u3058 1 \u7968\u3068\u3057\u3066\u6271\u3046\u3002\u7d50\u679c\u3001\u300c\u8efd\u5fae\u306a\u30a8\u30e9\u30fc\u304c\u591a\u3044\u6642\u9593\u5e2f\u306e 1 \u5206\u300d\u304c\u4e88\u7b97\u3092\u5927\u304d\u304f\u524a\u308b\u4e00\u65b9\u3001\u6df1\u591c\u306e\u5927\u898f\u6a21\u969c\u5bb3\u304c\u53cd\u6620\u3055\u308c\u306b\u304f\u3044\u9006\u8ee2\u73fe\u8c61\u304c\u751f\u3058\u308b\u3002\u30a4\u30d9\u30f3\u30c8\u30d9\u30fc\u30b9\u96c6\u8a08\u306b\u5207\u308a\u66ff\u3048\u308b\u3068\u6df1\u523b\u5ea6\u306b\u6bd4\u4f8b\u3057\u305f\u6d88\u8cbb\u306b\u306a\u308a\u3001\u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u304c\u300c\u4fe1\u3058\u3089\u308c\u308b\u30b7\u30b0\u30ca\u30eb\u300d\u3068\u3057\u3066\u6a5f\u80fd\u3059\u308b [19]\u3002\n\n### 4.6 \u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u306e\u62e1\u5f35\u9069\u7528\n\nThomson\u30fbLaing\uff08Pivotal\uff09\u306f SLI/SLO/\u30dd\u30ea\u30b7\u30fc\u306e\u30e2\u30c7\u30eb\u304c\u30c9\u30e1\u30a4\u30f3\u975e\u4f9d\u5b58\u306e\u6c4e\u7528\u69cb\u9020\u3067\u3042\u308b\u3053\u3068\u3092\u793a\u3057\u305f\u3002\n\n- **\u8106\u5f31\u6027\u30d0\u30b8\u30a7\u30c3\u30c8**: SLI = \u4f9d\u5b58\u30d1\u30c3\u30c1\u30ea\u30ea\u30fc\u30b9\u304b\u3089\u306e\u7d4c\u904e\u65e5\u6570\u3001SLO = \u6700\u5927\u8a31\u5bb9\u65e5\u6570\n- **\u30d5\u30a3\u30fc\u30c1\u30e3\u30fc\u30d5\u30ec\u30c3\u30b7\u30e5\u30cd\u30b9**: SLI = \u30d5\u30a3\u30fc\u30c1\u30e3\u30fc\u30ea\u30ea\u30fc\u30b9\u304b\u3089\u306e\u7d4c\u904e\u65e5\u6570\u3001SLO = \u30d6\u30ea\u30fc\u30c7\u30a3\u30f3\u30b0\u30a8\u30c3\u30b8\u5ea6\u306e\u7bc4\u56f2\n\nEquifax \u4fb5\u5bb3(CVE-2017-5638)\u306e\u4e8b\u4f8b\u3067\u300c30 \u65e5 SLO \u306e\u30d1\u30c3\u30c1\u9069\u7528\u30dd\u30ea\u30b7\u30fc\u304c\u3042\u308c\u3070 67 \u65e5\u5f8c\u306e\u4fb5\u5bb3\u306f\u9632\u3052\u305f\u300d\u3068\u5b9f\u8a3c\u3057\u305f [20]\u3002\n\n---\n\n## \u7b2c 5 \u7ae0\u3000SLO \u30d9\u30fc\u30b9\u30a2\u30e9\u30fc\u30c6\u30a3\u30f3\u30b0\n\n### 5.1 \u8a2d\u8a08\u539f\u5247\n\nSLO \u30d9\u30fc\u30b9\u30a2\u30e9\u30fc\u30c8\u306e\u76ee\u7684\u306f\u3001\u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u3092\u5927\u304d\u304f\u6d88\u8cbb\u3059\u308b\u91cd\u5927\u30a4\u30d9\u30f3\u30c8\u3092\u3001\u4e88\u7b97\u6d88\u8cbb\u524d\u306b\u901a\u77e5\u3059\u308b\u3053\u3068\u3067\u3042\u308b\u3002\u30a2\u30e9\u30fc\u30c8\u306f\u30a8\u30e9\u30fc\u7387\u306e\u7570\u5e38\u3067\u306f\u306a\u304f\u3001\u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u3078\u306e\u91cd\u5927\u306a\u8105\u5a01\u306b\u5bfe\u3057\u3066\u51fa\u3059 [21]\u3002\n\n### 5.2 \u8a55\u4fa1\u8ef8\n\n- **\u7cbe\u5ea6**: \u691c\u77e5\u3055\u308c\u305f\u30a4\u30d9\u30f3\u30c8\u306e\u3046\u3061\u91cd\u5927\u30a4\u30d9\u30f3\u30c8\u3067\u3042\u308b\u5272\u5408\n- **\u518d\u73fe\u7387**: \u91cd\u5927\u30a4\u30d9\u30f3\u30c8\u306e\u3046\u3061\u30a2\u30e9\u30fc\u30c8\u3055\u308c\u305f\u5272\u5408\n- **\u691c\u77e5\u6642\u9593**: \u901a\u77e5\u307e\u3067\u306e\u6642\u9593\n- **\u30ea\u30bb\u30c3\u30c8\u6642\u9593**: \u89e3\u6c7a\u5f8c\u306b\u30a2\u30e9\u30fc\u30c8\u304c\u6b8b\u308b\u6642\u9593\n\n### 5.3 \u30d0\u30fc\u30f3\u30ec\u30fc\u30c8\n\n\u30d0\u30fc\u30f3\u30ec\u30fc\u30c8\u306f SLO \u306b\u5bfe\u3057\u3066\u3069\u308c\u3060\u3051\u901f\u304f\u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u3092\u6d88\u8cbb\u3057\u3066\u3044\u308b\u304b\u3092\u8868\u3059\u3002\n\n| \u30d0\u30fc\u30f3\u30ec\u30fc\u30c8 | \u30a8\u30e9\u30fc\u7387 | 30 \u65e5\u30fb99.9% SLO \u3067\u306e\u67af\u6e07\u4e88\u6e2c |\n|---|---|---|\n| 1 | 0.1% | 30 \u65e5 |\n| 2 | 0.2% | 15 \u65e5 |\n| 10 | 1% | 3 \u65e5 |\n| 1000 | 100% | 43 \u5206 |\n\n[22]\n\n### 5.4 \u63a8\u5968\u69cb\u6210: \u8907\u6570\u30a6\u30a3\u30f3\u30c9\u30a6\u30fb\u8907\u6570\u30d0\u30fc\u30f3\u30ec\u30fc\u30c8\n\n\u63a8\u5968\u958b\u59cb\u70b9:\n\n- **\u30da\u30fc\u30b8\u7528**: 1 \u6642\u9593 2% \u4e88\u7b97\u6d88\u8cbb + 6 \u6642\u9593 5% \u4e88\u7b97\u6d88\u8cbb\n- **\u30c1\u30b1\u30c3\u30c8\u7528**: 3 \u65e5 10% \u4e88\u7b97\u6d88\u8cbb\n\n\u77ed\u7a93\u3092\u9577\u7a93\u306e 1/12 \u7a0b\u5ea6\u306b\u7f6e\u304d\u3001\u9577\u7a93\u3068\u77ed\u7a93\u304c\u540c\u6642\u306b\u95be\u5024\u3092\u8d85\u3048\u308b\u3068\u304d\u3060\u3051\u901a\u77e5\u3059\u308b\u3068\u3001\u7cbe\u5ea6\u3068\u30ea\u30bb\u30c3\u30c8\u6642\u9593\u306e\u30d0\u30e9\u30f3\u30b9\u304c\u6539\u5584\u3059\u308b [21]\u3002\n\nWilkinson\uff08Google\uff09\u306f Prometheus \u3067 `delta(errors[1h]) > (expected_events * error_budget / burn_period)` \u3092 SLO Fast Burn \u30a2\u30e9\u30fc\u30c8\u5f0f\u3068\u3057\u3066\u793a\u3057\u305f [23]\u3002\n\n### 5.5 \u30d0\u30fc\u30f3\u30ec\u30fc\u30c8\u30b0\u30e9\u30d5\u306e\u8aad\u307f\u65b9\n\nEBR \u30b0\u30e9\u30d5\u4e0a\u3067\u8b58\u5225\u3059\u3079\u304d 4 \u30d1\u30bf\u30fc\u30f3:\n\n1. **Slow burn**: \u5909\u66f4\u8d77\u56e0\u306e\u3058\u308f\u3058\u308f\u6d88\u8cbb\n2. **Fast burn**: \u5927\u898f\u6a21\u30a4\u30f3\u30b7\u30c7\u30f3\u30c8\u306b\u3088\u308b\u6025\u6fc0\u6d88\u8cbb\n3. **Recovery**: \u30d0\u30b0\u4fee\u6b63\u5f8c\u306e\u56de\u5fa9\n4. **No Incident drain**: \u30a4\u30f3\u30b7\u30c7\u30f3\u30c8\u306a\u3057\u306e\u7de9\u6162\u6d88\u8cbb(\u30d0\u30b0\u8d77\u56e0\u306e\u9045\u5ef6\u52a3\u5316\u3002\u30d0\u30fc\u30f3\u30ec\u30fc\u30c8\u30a2\u30e9\u30fc\u30c8\u3060\u3051\u3067\u306f\u6355\u6349\u56f0\u96e3)\n\n[19]\n\n### 5.6 \u4f4e\u30c8\u30e9\u30d5\u30a3\u30c3\u30af\u30fb\u6975\u7aef\u306a\u53ef\u7528\u6027\u76ee\u6a19\n\n- **\u4f4e\u30c8\u30e9\u30d5\u30a3\u30c3\u30af**: 10 \u30ea\u30af\u30a8\u30b9\u30c8/\u6642\u3067\u306f 1 \u5931\u6557\u304c 10% \u30a8\u30e9\u30fc\u7387\u3002\u5bfe\u7b56\u306f\u4eba\u5de5\u30c8\u30e9\u30d5\u30a3\u30c3\u30af\u3001\u95a2\u9023\u30b5\u30fc\u30d3\u30b9\u306e\u4e0a\u4f4d\u96c6\u7d04\u3001SLO \u518d\u4ea4\u6e09\n- **99.999% \u76ee\u6a19**: 100% \u969c\u5bb3\u304c 26 \u79d2\u3067\u4e88\u7b97\u67af\u6e07\u3002\u30a2\u30e9\u30fc\u30c8\u3067\u306f\u5b88\u308c\u306a\u3044\u305f\u3081\u3001\u30ab\u30ca\u30ea\u30a2\u30ea\u30ea\u30fc\u30b9\u306a\u3069\u8a2d\u8a08\u3067\u5168\u505c\u6b62\u78ba\u7387\u3092\u4e0b\u3052\u308b\n\n[21]\n\n### 5.7 SLO \u30a2\u30e9\u30fc\u30c8\u767a\u706b\u5f8c\u306e\u8a2d\u8a08\n\nSLO \u30a2\u30e9\u30fc\u30c8\u306f\u300c\u9cf4\u3089\u3059\u6761\u4ef6\u300d\u3060\u3051\u3067\u306a\u304f\u300c\u9cf4\u3063\u305f\u5f8c\u306b\u4f55\u3092\u898b\u308c\u3070\u3088\u3044\u304b\u300d\u3092\u542b\u3081\u3066\u8a2d\u8a08\u3055\u308c\u308b\u3079\u304d\u3067\u3042\u308b\u3002\u6c60\u7530\u5c06\u58eb\uff08\u9762\u767d\u6cd5\u4eba\u30ab\u30e4\u30c3\u30af\uff09\u306f SRE NEXT 2023 \u3067\u3001Warning \u30a2\u30e9\u30fc\u30c8\u306b\u5bfe\u3057\u5bfe\u8c61\u671f\u9593\u306e\u30ea\u30af\u30a8\u30b9\u30c8\u30d1\u30b9\u5225\u30ec\u30b9\u30dd\u30f3\u30b9\u30bf\u30a4\u30e0\u3092\u81ea\u52d5\u96c6\u8a08\u3059\u308b\u4ed5\u7d44\u307f\u3092\u793a\u3057\u305f [24]\u3002\n\n---\n\n## \u7b2c 6 \u7ae0\u3000\u53ef\u7528\u6027\u6307\u6a19\u306e\u9650\u754c\u3068\u9032\u5316\n\n### 6.1 \u30ca\u30a4\u30f3\u3060\u3051\u3067\u306f\u4e0d\u5341\u5206\n\nMogul+Wilkes\uff08Google\uff09\u306f\u300c\u30ca\u30a4\u30f3\u3060\u3051\u3067\u306f\u4e0d\u5341\u5206\u300d\u3068\u8ad6\u3058\u30013 \u3064\u306e\u9650\u754c\u3092\u6307\u6458\u3057\u305f\u3002\n\n- \u77ed\u6642\u9593\u65ad\u7d9a\u969c\u5bb3\u3068\u9577\u6642\u9593\u5927\u898f\u6a21\u969c\u5bb3\u3092\u533a\u5225\u3067\u304d\u306a\u3044\n- \u30b0\u30ec\u30fc\u30b9\u30d5\u30eb\u30c7\u30b0\u30e9\u30c7\u30fc\u30b7\u30e7\u30f3\u3092\u8a18\u8ff0\u3067\u304d\u306a\u3044\n- \u30d3\u30b8\u30cd\u30b9\u7684\u91cd\u8981\u65e5\u306e\u91cd\u307f\u4ed8\u3051\u304c\u3067\u304d\u306a\u3044\n\n[5]\n\n### 6.2 \u826f\u3044\u53ef\u7528\u6027\u30e1\u30c8\u30ea\u30af\u30b9\u306e\u4e09\u8981\u4ef6\n\nHauer+\uff08Google\uff09\u306f\u300c\u826f\u3044\u53ef\u7528\u6027\u30e1\u30c8\u30ea\u30af\u30b9\u306f**\u6709\u610f\u7fa9\u6027**(\u30e6\u30fc\u30b6\u30fc\u4f53\u9a13\u3092\u6349\u3048\u308b)\u30fb**\u6bd4\u4f8b\u6027**(\u5909\u5316\u306b\u6bd4\u4f8b\u3059\u308b)\u30fb**\u5b9f\u7528\u6027**(\u539f\u56e0\u306e\u6d1e\u5bdf\u3092\u4e0e\u3048\u308b)\u306e\u4e09\u8981\u4ef6\u3092\u540c\u6642\u306b\u6e80\u305f\u3059\u3079\u304d\u300d\u3068\u5b9a\u5f0f\u5316\u3057\u305f [25]\u3002\n\n\u65e2\u5b58\u6307\u6a19\u306e\u9650\u754c:\n\n- **\u6210\u529f\u7387**: \u6700\u6d3b\u767a\u30e6\u30fc\u30b6\u30fc\u306b\u6700\u5927 1,000 \u500d\u504f\u308a\u3001\u969c\u5bb3\u4e2d\u306e\u30ea\u30af\u30a8\u30b9\u30c8\u65ad\u5ff5\u3092\u904e\u5c11\u8a55\u4fa1\n- **\u30a4\u30f3\u30b7\u30c7\u30f3\u30c8\u6bd4**: \u4efb\u610f\u306e\u300c\u30a4\u30f3\u30b7\u30c7\u30f3\u30c8\u300d\u3057\u304d\u3044\u5024\u306b\u4f9d\u5b58\u3057\u3001\u5927\u898f\u6a21\u5206\u6563\u30b7\u30b9\u30c6\u30e0\u306b\u4e0d\u9069\u5207\n- **\u5408\u6210\u30d7\u30ed\u30fc\u30d6**: \u30e6\u30fc\u30b6\u30fc\u306e\u5b9f\u969b\u306e\u30ef\u30fc\u30af\u30ed\u30fc\u30c9\u3092\u4ee3\u8868\u3057\u306a\u3044\n\n### 6.3 \u30a6\u30a3\u30f3\u30c9\u30a6\u4ed8\u304d\u30e6\u30fc\u30b6\u30fc\u30a2\u30c3\u30d7\u30bf\u30a4\u30e0\n\nHauer+ \u304c Google G Suite \u3067\u8a55\u4fa1\u30fb\u672c\u756a\u5c55\u958b\u3057\u305f\u65b0\u6307\u6a19:\n\n1. \u5404\u30e6\u30fc\u30b6\u30fc\u306e\u7d30\u7c92\u5ea6\u30ea\u30af\u30a8\u30b9\u30c8\u30ed\u30b0\u304b\u3089\u30a2\u30c3\u30d7\u5206\u30fb\u30c0\u30a6\u30f3\u5206\u3092\u7b97\u51fa\u3057\u3001\u5168\u30e6\u30fc\u30b6\u30fc\u306e\u5747\u7b49\u52a0\u91cd\u5e73\u5747\u3067\u96c6\u7d04\n2. 1 \u5206\u304b\u3089\u56db\u534a\u671f\u307e\u3067\u3059\u3079\u3066\u306e\u30a6\u30a3\u30f3\u30c9\u30a6\u30b5\u30a4\u30ba\u3067\u6700\u60aa\u53ef\u7528\u6027\u3092\u7b97\u51fa\u3057\u3001MCR(Maximum Contiguous Ratio)\u66f2\u7dda\u3068\u3057\u3066\u53ef\u8996\u5316\u3002\u300c\u819d\u300d\u306e\u4f4d\u7f6e\u3067\u77ed\u6642\u9593\u65ad\u7d9a\u969c\u5bb3\u3068\u9577\u6642\u9593\u5927\u898f\u6a21\u969c\u5bb3\u3092\u533a\u5225\n\n[25]\n\n### 6.4 SLE/CBE \u2014 \u6cd5\u5f8b\u5bb6\u7684\u601d\u8003\u304b\u3089\u7d71\u8a08\u5bb6\u7684\u601d\u8003\u3078\n\nMogul+Wilkes \u306f SLO \u5b9a\u7fa9\u306e\u56f0\u96e3\u3055\u3092\u7d71\u8a08\u5b66\u7684\u610f\u601d\u6c7a\u5b9a\u3068\u306e\u540c\u578b\u6027\u3068\u3057\u3066\u6349\u3048\u3001\u8ee2\u63db\u3092\u63d0\u5531\u3057\u305f\u3002\n\n- **SLE(\u30b5\u30fc\u30d3\u30b9\u30ec\u30d9\u30eb\u671f\u5f85)**: \u30d7\u30ed\u30d0\u30a4\u30c0\u304c\u901a\u5e38\u6761\u4ef6\u4e0b\u3067\u9867\u5ba2\u306b\u671f\u5f85\u3055\u305b\u308b\u6319\u52d5\u3002\u7d50\u679c\u4fdd\u8a3c\u3067\u306f\u306a\u304f\u671f\u5f85\u7ba1\u7406\n- **CBE(\u9867\u5ba2\u6319\u52d5\u671f\u5f85)**: \u30d7\u30ed\u30d0\u30a4\u30c0\u304c SLE \u3092\u6e80\u305f\u3059\u305f\u3081\u306b\u524d\u63d0\u3068\u3059\u308b\u9867\u5ba2\u5074\u306e\u6319\u52d5\n\n[5]\n\n### 6.5 \u96c6\u8a08\u304c\u96a0\u3059\u554f\u984c\n\nSLO \u30e2\u30c7\u30eb\u306f\u6642\u9593\u30fb\u7a7a\u9593\u3067\u7dda\u5f62\u6027\u3092\u4eee\u5b9a\u3057\u3066\u304a\u308a\u3001\u96c6\u7d04\u306b\u3088\u3063\u3066\u6df1\u523b\u306a\u554f\u984c\u3092\u96a0\u853d\u3059\u308b\u3002\n\n- **\u6642\u9593\u7dda\u5f62\u6027**: 1,000\u00d71 \u5206\u505c\u6b62 = 1\u00d71,000 \u5206\u505c\u6b62\u3068\u6271\u308f\u308c\u308b\u304c\u3001\u30e6\u30fc\u30b6\u30fc\u5f71\u97ff\u306f\u5168\u304f\u7570\u306a\u308b\n- **\u7a7a\u9593\u7dda\u5f62\u6027**: \u4e00\u90e8\u30e6\u30fc\u30b6\u30fc\u3078\u306e\u5b8c\u5168\u969c\u5bb3 = \u5168\u30e6\u30fc\u30b6\u30fc\u3078\u306e\u8efd\u5fae\u306a\u969c\u5bb3\u3068\u6271\u308f\u308c\u308b\n\n[26]\n\nPalcuie\uff08Google\uff09\u306f\u300c\u30d7\u30ed\u30d0\u30a4\u30c0\u306e\u96c6\u8a08 SLO \u304c 99.999% \u3092\u4fdd\u3064\u4e00\u65b9\u3067\u500b\u5225\u9867\u5ba2\u304c\u5b8c\u5168\u306a\u969c\u5bb3\u72b6\u614b\u306b\u306a\u308a\u3046\u308b\u300d\u3053\u3068\u3092 GCE \u306e\u5b9f\u30c7\u30fc\u30bf\u3067\u793a\u3057\u3001\u9867\u5ba2\u5358\u4f4d SLO \u3078\u306e\u79fb\u884c\u3092\u63a8\u9032\u3057\u305f [27]\u3002\n\n### 6.6 \u300c5 \u30a8\u30e9\u30fc\u306e\u30eb\u30fc\u30eb\u300d\u306b\u3088\u308b\u52d5\u7684\u30bf\u30fc\u30b2\u30c3\u30c8\n\n\u5c11\u30c8\u30e9\u30d5\u30a3\u30c3\u30af\u9867\u5ba2\u3078\u306e\u4e00\u5f8b SLO \u9069\u7528\u306f\u73fe\u5b9f\u7684\u3067\u306a\u3044\u3002Palcuie \u306f\u9867\u5ba2\u5358\u4f4d SLO \u306e\u30bf\u30fc\u30b2\u30c3\u30c8\u3092\u30c8\u30e9\u30d5\u30a3\u30c3\u30af\u91cf\u306b\u5fdc\u3058\u3066 `target = 1 - 5 / total_requests` \u3068\u52d5\u7684\u8a2d\u5b9a\u3059\u308b\u624b\u6cd5\u3092\u793a\u3057\u305f\u30021 \u4e07\u30ea\u30af\u30a8\u30b9\u30c8\u4ee5\u4e0a\u3067 99.95% \u306b\u53ce\u675f\u3059\u308b\u304c\u300110 \u30ea\u30af\u30a8\u30b9\u30c8\u3067\u306f 50% \u3068\u306a\u308b [27]\u3002\n\n---\n\n## \u7b2c 7 \u7ae0\u3000SLO \u306e\u7d44\u7e54\u7684\u5c0e\u5165 \u2014 \u6280\u8853\u304b\u3089\u6587\u5316\u3078\n\n### 7.1 \u6bb5\u968e\u7684\u5c0e\u5165\u304c\u9375\n\nSLO \u5c0e\u5165\u306e\u5931\u6557\u8981\u56e0\u304c\u300c\u6280\u8853\u300d\u3088\u308a\u300c\u7d44\u7e54\u30fb\u4eba\u30fb\u30d7\u30ed\u30bb\u30b9\u300d\u3067\u3042\u308b\u3053\u3068\u306f\u8907\u6570\u306e\u30bd\u30fc\u30b9\u3067\u72ec\u7acb\u306b\u78ba\u8a8d\u3055\u308c\u3066\u3044\u308b [28][29]\u3002\n\nTakamura\uff08Topotal\uff09\u306f SLO \u9055\u53cd\u30dd\u30ea\u30b7\u30fc\u306e\u6bb5\u968e\u7684\u62e1\u5927\u3092\u4f53\u7cfb\u5316\u3057\u305f\u3002\n\n| \u30ec\u30d9\u30eb | \u30a2\u30af\u30b7\u30e7\u30f3 |\n|---|---|\n| Lv1 | \u306a\u306b\u3082\u3057\u306a\u3044(\u8a08\u6e2c\u3068\u53ef\u8996\u5316\u306e\u307f) |\n| Lv2 | \u8abf\u67fb\u5224\u65ad\u306e\u30c8\u30ea\u30ac\u30fc\u3068\u3059\u308b |\n| Lv3 | \u30c1\u30fc\u30e0\u5185\u3067\u30a2\u30af\u30b7\u30e7\u30f3\u5b9f\u884c |\n| Lv4 | \u4ed6\u30c1\u30fc\u30e0\u3068\u306e\u8abf\u6574\u3092\u542b\u3080\u30a2\u30af\u30b7\u30e7\u30f3 |\n| Lv5 | \u958b\u767a\u30b9\u30b1\u30b8\u30e5\u30fc\u30eb\u306e\u610f\u601d\u6c7a\u5b9a\u306b SLO \u3092\u7d44\u307f\u8fbc\u3080 |\n\n[29]\n\n### 7.2 \u300c\u4eee\u5024\u304b\u3089\u59cb\u3081\u3066\u30d5\u30a3\u30fc\u30c9\u30d0\u30c3\u30af\u3067\u6d17\u7df4\u3059\u308b\u300d\n\n4 \u540d\u306e\u72ec\u7acb\u3057\u305f\u5b9f\u8df5\u8005\u304c\u5171\u901a\u3057\u3066\u300c\u5b8c\u74a7\u306a\u5024\u3092\u8ffd\u3044\u6c42\u3081\u305a\u52d5\u304b\u3057\u306a\u304c\u3089\u4fee\u6b63\u3059\u308b\u300d\u3092\u6210\u529f\u306e\u8d77\u70b9\u3068\u3057\u3066\u9078\u3093\u3067\u3044\u308b\u3002\n\n- \u6e21\u8fba\u8d77(Mackerel, SRE NEXT 2023):\u300cSLI \u3068\u4eee\u5024\u3092\u6c7a\u3081\u3066\u898b\u76f4\u3057\u30d5\u30ed\u30fc\u3092\u4f5c\u308a\u3001\u307e\u305a\u3068\u308a\u3042\u3048\u305a\u59cb\u3081\u305f\u300d\n- Takamura(Topotal, 2026 \u795e\u6238):\u300c\u5c0f\u3055\u304f\u59cb\u3081\u308b\u300d4 \u30b9\u30c6\u30c3\u30d7\u306e\u4f53\u7cfb\u5316\n- Durst(Spring Health, SREcon25 EMEA): 4 \u524d\u63d0\u6761\u4ef6\u78ba\u8a8d\u5f8c\u306e\u6bb5\u968e\u7684\u5c0e\u5165\n- Hidalgo(Squarespace, SREcon19 EMEA):\u300c\u7fcc\u65e5\u969c\u5bb3\u3067\u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u67af\u6e07\u3057\u305f\u304c\u8a31\u53ef\u8a3c\u3068\u3057\u3066\u5373\u5ea7\u306b\u6a5f\u80fd\u3057\u305f\u300d\n\n[30][28]\n\n### 7.3 \u30b9\u30c6\u30fc\u30af\u30db\u30eb\u30c0\u30fc\u306e\u300c\u6697\u9806\u5fdc\u671f\u9593\u300d\n\nSmith\uff08Campspot\uff09\u306f SLO \u5c0e\u5165\u4e8b\u4f8b\u3067\u3001\u55b6\u696d\u30c1\u30fc\u30e0\u3092 SLO \u30ef\u30fc\u30af\u30b7\u30e7\u30c3\u30d7\u306b\u62db\u3044\u305f\u3068\u3053\u308d\u300cUptime 99.999% \u306e\u8a71\u3067\u306f\u306a\u3044\u306e\u304b\u300d\u3068\u6df7\u4e71\u304c\u751f\u3058\u305f\u3068\u5831\u544a\u3057\u305f\u3002\u300c\u53ef\u7528\u6027 = \u7a3c\u50cd\u7387\u300d\u304b\u3089\u300c\u826f\u3044\u4f53\u9a13\u306e\u5272\u5408\u300d\u3078\u306e\u8a8d\u77e5\u7684\u8ee2\u63db\u306b\u306f\u6e96\u5099\u671f\u9593\u304c\u5fc5\u8981\u3067\u3042\u308a\u3001SLO \u5b9f\u88c5\u306e\u524d\u306b\u975e\u6280\u8853\u30b9\u30c6\u30fc\u30af\u30db\u30eb\u30c0\u30fc\u3068\u306e\u300cAdjust\u300d\u30d5\u30a7\u30fc\u30ba\u3092\u8a2d\u3051\u308b\u3079\u304d\u3060\u3068\u3059\u308b [31]\u3002\n\n### 7.4 SLO \u306f\u4ea4\u6e09\u30d7\u30ed\u30bb\u30b9\u3067\u3042\u308b\n\nCoulter\uff08AppDynamics\uff09\u306f SLO \u3092\u6280\u8853\u7684\u306b\u300c\u6b63\u3057\u3044\u5024\u3092\u8a08\u7b97\u3059\u308b\u300d\u554f\u984c\u3067\u306f\u306a\u304f\u3001\u30b9\u30c6\u30fc\u30af\u30db\u30eb\u30c0\u30fc\u4ea4\u6e09\u3068\u3057\u3066\u6271\u3046\u8996\u70b9\u3092\u793a\u3057\u305f\u3002\u300cPrepare to Engage \u2192 Warmup \u2192 Test Drive \u2192 Assess \u2192 Propose \u2192 [RECUR] \u2192 Agree\u300d\u3068\u3044\u3046\u53cd\u5fa9\u7684\u4ea4\u6e09\u30d5\u30ed\u30fc\u3092\u63d0\u6848\u3057\u3001SLO \u306f\u4e00\u5ea6\u5408\u610f\u3059\u308c\u3070\u7d42\u308f\u308a\u3067\u306f\u306a\u304f\u7d99\u7d9a\u7684\u306b\u518d\u4ea4\u6e09\u3055\u308c\u308b\u3082\u306e\u3068\u3057\u3066\u8a2d\u8a08\u3055\u308c\u308b [14]\u3002\n\n### 7.5 \u7d44\u7e54\u7684\u524d\u63d0\u6761\u4ef6\n\n\u8fd1\u85e4\u6b66\u58eb(Recruit, SRE NEXT 2022)\u306f\u3001SLI/SLO \u306e\u5b9a\u7fa9\u30fb\u89b3\u5bdf\u6587\u5316\u3092 2 \u30d7\u30ed\u30c0\u30af\u30c8 15 \u30c1\u30fc\u30e0\u306b\u5c0e\u5165\u3057\u305f\u5f8c\u3082\u3001Error Budget Policy \u304c\u5b9a\u7740\u3057\u306a\u304b\u3063\u305f\u539f\u56e0\u3092\u300c**\u975e\u6a5f\u80fd\u8981\u6c42\u306b\u5bfe\u51e6\u3059\u308b\u4e88\u7b97\u30fb\u6a29\u9650\u304c\u958b\u767a\u30c1\u30fc\u30e0\u306b\u306a\u304b\u3063\u305f**\u300d\u3068\u5206\u6790\u3057\u305f\u3002\u6280\u8853\u6226\u7565\u30b0\u30eb\u30fc\u30d7\u306e\u767a\u8db3\u3068\u300c\u65b0\u898f:\u30a8\u30f3\u30cf\u30f3\u30b9:\u6280\u8853\u7684\u8ca0\u50b5\u89e3\u6d88 = 1:1:1\u300d\u3068\u3044\u3046\u4e88\u7b97\u5ba3\u8a00\u304c\u89e3\u6c7a\u3092\u3082\u305f\u3089\u3057\u305f [32]\u3002\n\nDurst(Spring Health, SREcon25 EMEA)\u306f\u5c0e\u5165\u306e 4 \u524d\u63d0\u6761\u4ef6\u3092\u793a\u3057\u305f: (1) \u6240\u6709\u6a29\u3001(2) \u6a19\u6e96\u30d7\u30ed\u30bb\u30b9\u3001(3) \u4fdd\u8b77\u6642\u9593\u3001(4) SLO \u8a08\u6e2c\u57fa\u76e4 [28]\u3002\n\n### 7.6 \u5b9f\u5c0e\u5165\u4e8b\u4f8b\u306e\u6bd4\u8f03\n\n| \u7d44\u7e54 | \u898f\u6a21 | \u30a2\u30d7\u30ed\u30fc\u30c1 | \u7d50\u679c |\n|---|---|---|---|\n| **Evernote** | 2.2 \u5104\u30e6\u30fc\u30b6\u30fc | \u5916\u5f62\u76e3\u8996\u30d9\u30fc\u30b9\u306e\u6708\u6b21 99.95% \u304b\u3089\u958b\u59cb\u3001Google CRE \u3068\u5354\u50cd | \u30ea\u30ea\u30fc\u30b9\u7a93 5\u21922 \u56de\u306b\u524a\u6e1b [12] |\n| **The Home Depot** | 2,200 \u8d85\u5e97\u8217 | VALET \u67a0\u7d44\u307f + \u81ea\u52d5\u53ce\u96c6\u57fa\u76e4 + \u7ba1\u7406\u8077\u30a4\u30f3\u30bb\u30f3\u30c6\u30a3\u30d6 | 0\u2192800 \u30b5\u30fc\u30d3\u30b9\u3092 1 \u5e74\u672a\u6e80\u3067\u5c55\u958b [12] |\n| **Atlassian** | \u5927\u4f01\u696d | Error Budgets 0.1(13 \u9031\u4e2d 7 \u9031\u672a\u9054\u3067\u30a2\u30af\u30b7\u30e7\u30f3)\u304b\u3089\u958b\u59cb | \u6bb5\u968e\u7684\u306b\u57fa\u6e96\u5f15\u304d\u7de0\u3081 [33] |\n| **Squarespace** | \u4e2d\u898f\u6a21 | Ceph Object Storage \u306b 6 \u30b9\u30c6\u30c3\u30d7\u3067 SLO \u5b9f\u88c5 | \u30d7\u30ed\u30fc\u30d0\u30fc\u8a08\u6e2c\u3067\u65b0\u898f\u30b5\u30fc\u30d3\u30b9\u306e SLI \u78ba\u7acb [34] |\n| **Spring Health** | 200 \u540d / 8 SRE | 3 \u5e74\u9593\u306e\u8a66\u884c\u932f\u8aa4\u5f8c\u306b\u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u8d77\u70b9\u306e\u30b3\u30fc\u30c9\u30d5\u30ea\u30fc\u30ba RFC \u7d4c\u55b6\u627f\u8a8d | \u30b9\u30bf\u30fc\u30c8\u30a2\u30c3\u30d7\u3067\u3082 Lv5 \u5230\u9054\u53ef\u80fd\u3092\u5b9f\u8a3c [28] |\n| **Riot Games** | \u30b2\u30fc\u30e0 | Player Minutes(CCU \u91cd\u307f\u4ed8\u304d)+ CEO OKR \u5f37\u5236 | \u6587\u5316\u5909\u9769\u3092 3 \u5c64\u3067\u63a8\u9032 [35] |\n| **Ant Group** | 60+ K8s \u30af\u30e9\u30b9\u30bf | GitOps + K8s \u5ba3\u8a00\u7684\u7ba1\u7406\u3067 SLO \u7206\u767a\u554f\u984c\u3092\u5438\u53ce | 99.999% \u76ee\u6a19\u3092\u904b\u7528 [36] |\n\n---\n\n## \u7b2c 8 \u7ae0\u3000\u5fdc\u7528\u3068\u62e1\u5f35\n\n### 8.1 \u30c7\u30fc\u30bf\u96c6\u7d04\u578b\u30b5\u30fc\u30d3\u30b9\u306e SLO\n\nBooking.com \u3067\u306f\u53ef\u7528\u6027\u30fb\u30ec\u30a4\u30c6\u30f3\u30b7 SLO \u3060\u3051\u3067\u306f\u30b9\u30c6\u30fc\u30af\u30db\u30eb\u30c0\u30fc\u304c\u95a2\u5fc3\u3092\u6301\u305f\u306a\u304b\u3063\u305f\u3002**\u30c7\u30fc\u30bf\u306e\u4e00\u8cab\u6027\u30fb\u65b0\u9bae\u6027\u30fb\u5b8c\u5168\u6027\u30fb\u8010\u4e45\u6027**\u3092\u300c\u30c7\u30fc\u30bf\u54c1\u8cea SLO\u300d\u3068\u3057\u3066\u8a08\u6e2c\u30fb\u76ee\u6a19\u5316\u3059\u308b\u307e\u3067\u306f SLO \u304c\u610f\u601d\u6c7a\u5b9a\u306b\u4f7f\u308f\u308c\u306a\u304b\u3063\u305f\u3002SLO \u304c\u660e\u78ba\u306b\u5b9a\u7fa9\u3055\u308c\u308b\u3068\u3001Freshness Probe \u304c\u30c8\u30e9\u30d5\u30a3\u30c3\u30af\u81ea\u52d5\u505c\u6b62(\u81ea\u52d5\u7de9\u548c)\u3001Completeness Probe \u304c Hadoop \u30b9\u30ca\u30c3\u30d7\u30b7\u30e7\u30c3\u30c8\u304b\u3089\u306e\u518d\u51e6\u7406(\u81ea\u52d5\u4fee\u5fa9)\u306e\u30c8\u30ea\u30ac\u30fc\u3068\u3057\u3066\u6a5f\u80fd\u3057\u305f [37]\u3002\n\n### 8.2 \u975e\u540c\u671f\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3\n\neBay \u306f\u53ef\u7528\u6027 SLI(SUCCESS / ABANDONED \u6bd4\u7387)\u3068\u30ec\u30a4\u30c6\u30f3\u30b7 SLI(\u7d2f\u7a4d end-to-end histogram)\u306e 2 \u7a2e\u985e\u3067\u3001\u975e\u540c\u671f\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3\u306e Freshness\u30fbQuality\u30fbThroughput \u3092\u4ee3\u66ff\u3067\u304d\u308b\u3053\u3068\u3092\u5b9f\u88c5\u3067\u793a\u3057\u305f\u3002\u30d0\u30fc\u30f3\u30ec\u30fc\u30c8\u30a2\u30e9\u30fc\u30c8\u306b\u52a0\u3048\u3066 `absent(sum(sli_valid_events{...}))` \u3067\u30e1\u30c8\u30ea\u30af\u30b9\u6b20\u640d\u3092\u691c\u77e5\u3059\u308b\u30c7\u30fc\u30bf\u640d\u5931\u30a2\u30e9\u30fc\u30c8\u3092\u4f75\u7f6e\u3059\u308b\u5fc5\u8981\u304c\u3042\u308b [38]\u3002\n\n### 8.3 SLO \u62e1\u6563(\u9ad8\u30ec\u30d9\u30eb\u2192\u4f4e\u30ec\u30d9\u30eb\u5206\u89e3)\n\nSedlak+\uff08TU Wien\uff09\u306f\u30de\u30a4\u30af\u30ed\u30b5\u30fc\u30d3\u30b9\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3\u306b\u304a\u3044\u3066\u9ad8\u30ec\u30d9\u30eb SLO \u3092\u81ea\u52d5\u7684\u306b\u4f4e\u30ec\u30d9\u30eb SLO \u3078\u5206\u89e3\u3059\u308b 3 \u30b9\u30c6\u30c3\u30d7\u65b9\u6cd5\u8ad6\u3092\u63d0\u6848\u3057\u305f: (1) \u30d9\u30a4\u30b8\u30a2\u30f3\u30cd\u30c3\u30c8\u30ef\u30fc\u30af\u5b66\u7fd2(BNL)\u306b\u3088\u308b\u30b5\u30fc\u30d3\u30b9\u5909\u6570\u9593\u306e\u6761\u4ef6\u4ed8\u304d\u4f9d\u5b58\u30b0\u30e9\u30d5\u69cb\u7bc9\u3001(2) \u9ad8\u30ec\u30d9\u30eb SLO \u5909\u6570\u304b\u3089\u30b0\u30e9\u30d5\u3092\u9061\u308a\u5909\u6570\u6d88\u53bb\u6cd5\u3067\u8a31\u5bb9\u7bc4\u56f2\u3092\u63a8\u5b9a\u3059\u308b\u62e1\u6563(HLD)\u3001(3) \u91cd\u8907\u5236\u7d04\u306e\u77db\u76fe\u691c\u77e5\u3068\u81ea\u5f8b\u7684\u89e3\u6c7a(CIR)\u3002\u8a55\u4fa1\u3067\u306f VehicleRouting\u30fbTrafficPrediction\u30fbLiveMonitoring \u306e 3 \u30d1\u30a4\u30d7\u30e9\u30a4\u30f3\u3067\u9ad8\u30ec\u30d9\u30eb SLO \u9054\u6210\u7387 83\u301c100% \u3092\u9054\u6210\u3057\u305f\u3002\n\n\u5236\u7d04\u306e\u53b3\u3057\u3055\u3092\u6c7a\u3081\u308b\u8d85\u30d1\u30e9\u30e1\u30fc\u30bf **\u03bb** \u306b\u306f\u91cd\u8981\u306a\u30c8\u30ec\u30fc\u30c9\u30aa\u30d5\u304c\u78ba\u8a8d\u3055\u308c\u305f\u3002\u03bb \u3092\u5927\u304d\u304f\u3059\u308b(\u8a31\u5bb9\u7bc4\u56f2\u3092\u72ed\u3081\u308b)\u3068\u9ad8\u30ec\u30d9\u30eb SLO \u9054\u6210\u7387\u306f\u5411\u4e0a\u3059\u308b\u304c\u3001\u904e\u5ea6\u306b\u5927\u304d\u304f\u3059\u308b\u3068\u4f4e\u30ec\u30d9\u30eb SLO \u9593\u3067\u30b3\u30f3\u30d5\u30ea\u30af\u30c8\u304c\u767a\u751f\u3057\u9054\u6210\u7387\u304c 0 \u3078\u6025\u843d\u3059\u308b\u3002\u30b3\u30f3\u30d5\u30ea\u30af\u30c8\u306f\u7a4d\u96c6\u5408\u304c\u7a7a\u3067\u306a\u3044**\u30de\u30a4\u30ca\u30fc\u30b3\u30f3\u30d5\u30ea\u30af\u30c8**(\u4ea4\u5dee\u533a\u9593\u3092\u63a1\u7528\u3057\u3066\u81ea\u52d5\u89e3\u6c7a)\u3068\u3001\u4e92\u3044\u306b\u7d20\u306a**\u30e1\u30b8\u30e3\u30fc\u30b3\u30f3\u30d5\u30ea\u30af\u30c8**(\u89e3\u6c7a\u4e0d\u80fd\u3068\u3057\u3066\u958b\u767a\u8005\u3078\u901a\u77e5)\u306e 2 \u7a2e\u306b\u5206\u985e\u3055\u308c\u308b\u3002\u4f4e\u30ec\u30d9\u30eb SLO \u5145\u8db3\u7387\u306f\u5e38\u306b\u9ad8\u30ec\u30d9\u30eb SLO \u5145\u8db3\u7387\u3088\u308a\u9ad8\u304f\u3001\u9ad8\u30ec\u30d9\u30eb\u5145\u8db3\u306f\u4f4e\u30ec\u30d9\u30eb\u5145\u8db3\u306e\u5e30\u7d50\u3068\u3057\u3066\u5f97\u3089\u308c\u308b\u3053\u3068\u3082\u78ba\u8a8d\u3055\u308c\u305f [39]\u3002\n\n### 8.4 SLX \u2014 \u8abf\u67fb\u306e\u305f\u3081\u306e\u62e1\u5f35\u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af\n\nSLO \u306f\u691c\u77e5\u306b\u306f\u5f37\u3044\u304c\u8abf\u67fb\u306b\u306f\u5411\u304b\u306a\u3044\u3002Ding\u30fbZhang\uff08Ant Group\uff09\u306f\u88dc\u5b8c\u6982\u5ff5\u3068\u3057\u3066 SLF(SLI \u3092\u8a73\u7d30\u30e9\u30d9\u30eb\u6b21\u5143\u3067\u30b9\u30e9\u30a4\u30b9)\u3068 SLD(\u4f9d\u5b58\u30b5\u30fc\u30d3\u30b9\u306e\u30e1\u30c8\u30ea\u30af\u30b9)\u3092\u8ffd\u52a0\u3057\u3001SLX Graph \u3092\u30b0\u30e9\u30d5\u8d70\u67fb\u3059\u308b\u3053\u3068\u3067\u7570\u5e38 SLO \u4f9d\u5b58\u30c1\u30a7\u30fc\u30f3\u3092\u81ea\u52d5\u7d5e\u308a\u8fbc\u3080 [36]\u3002\n\n### 8.5 \u30cf\u30fc\u30c9\u30b7\u30e3\u30fc\u30c9\u30b7\u30b9\u30c6\u30e0\n\n\u6c34\u5e73\u30b9\u30b1\u30fc\u30eb\u3067\u306f 3 \u30ce\u30fc\u30c9\u4e2d 1 \u30ce\u30fc\u30c9\u969c\u5bb3 = \u5168\u4f53 SLO 66%(\u6bd4\u4f8b\u53cd\u6620)\u3060\u304c\u3001\u30cf\u30fc\u30c9\u30b7\u30e3\u30fc\u30c9\u3067\u306f SLO 0%/100%/100% \u304c\u5168\u4f53\u96c6\u8a08\u3067\u306f 66% \u3068\u8aa4\u3063\u3066\u5065\u5168\u306b\u898b\u3048\u308b\u3002**\u30cf\u30fc\u30c9\u30b7\u30e3\u30fc\u30c9\u3067\u306f\u8ad6\u7406\u30a4\u30f3\u30b9\u30bf\u30f3\u30b9\u3054\u3068\u306b\u72ec\u7acb\u3057\u305f SLO \u304c\u4e0d\u53ef\u6b20** [9]\u3002\n\n### 8.6 HPC \u3078\u306e\u9069\u5fdc\n\nLANL \u306e Lueninghoener \u306f HPC \u74b0\u5883\u3067\u300c\u5404\u30af\u30e9\u30b9\u30bf\u56db\u534a\u671f 30 \u6642\u9593\u300d\u3092\u30c0\u30a6\u30f3\u30bf\u30a4\u30e0\u4e88\u7b97\u3068\u5b9a\u7fa9\u3057\u3001\u30d0\u30fc\u30f3\u30c0\u30a6\u30f3\u30c1\u30e3\u30fc\u30c8\u3067\u53ef\u8996\u5316\u3057\u305f\u3002\u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u306e\u672c\u8cea\u306f\u300c\u5358\u4f4d(\u30ea\u30af\u30a8\u30b9\u30c8\u5931\u6557)\u300d\u3067\u306f\u306a\u304f\u300c\u30ea\u30bd\u30fc\u30b9\u3092\u8ffd\u8de1\u3057\u3066\u610f\u601d\u6c7a\u5b9a\u306b\u4f7f\u3046\u300d\u69cb\u9020\u306b\u3042\u308b [40]\u3002\n\n### 8.7 IoT\u30fb\u30e2\u30d3\u30ea\u30c6\u30a3\n\nLuup \u3067\u306f\u96fb\u52d5\u30ad\u30c3\u30af\u30dc\u30fc\u30c9\u30fb\u96fb\u52d5\u30a2\u30b7\u30b9\u30c8\u81ea\u8ee2\u8eca\u3092\u5bfe\u8c61\u306b\u3001CUJ \u306e\u4ee3\u308f\u308a\u306b CMC(Critical Machine Communication)\u3092 SLI \u8a2d\u8a08\u306e\u8d77\u70b9\u3068\u3057\u305f\u3002\u8a08\u6e2c\u5bfe\u8c61\u306f\u30c7\u30d0\u30a4\u30b9\u306e\u5b9a\u671f\u901a\u4fe1\u5fdc\u7b54\u3084\u901a\u4fe1\u9014\u7d76\u3067\u3042\u308a\u3001SLI \u304c\u30ea\u30af\u30a8\u30b9\u30c8\u99c6\u52d5\u304b\u3089\u7269\u7406\u30c7\u30d0\u30a4\u30b9\u901a\u4fe1\u3078\u62e1\u5f35\u3055\u308c\u305f\u4e8b\u4f8b [41]\u3002\n\nLuup \u306f 2024 \u5e74\u3001Enabling SLO \u306e\u5bfe\u8c61\u3092 iOS/Android \u30af\u30e9\u30a4\u30a2\u30f3\u30c8\u3078\u62e1\u5f35\u3057\u305f\u3002API \u306e\u307f\u306e\u8a08\u6e2c\u3067\u306f BLE \u901a\u4fe1(\u65bd\u9320\u30fb\u89e3\u9320\u30fb\u30e9\u30a4\u30c9\u958b\u59cb\u7d42\u4e86)\u3084 Firestore \u3078\u306e\u76f4\u63a5\u30a2\u30af\u30bb\u30b9\u304c SLI \u304b\u3089\u6f0f\u308c\u308b\u305f\u3081\u3001PdM\u30fbSWE\u30fbSRE \u306e\u4e09\u8005\u5354\u50cd\u3067\u30e6\u30fc\u30b6\u30fc\u30b8\u30e3\u30fc\u30cb\u30fc\u3092\u518d\u8a2d\u5b9a\u3057\u3001CUJ \u5358\u4f4d\u306e\u30de\u30c8\u30ea\u30af\u30b9\u306b\u843d\u3068\u3057\u8fbc\u3093\u3060\u3002Latency SLI \u306b\u306f Core Web Vitals \u306e LCP Good Score \u304c\u63a1\u7528\u3059\u308b **p75** \u3092\u7528\u3044\u3001Datadog \u306e Time Slice SLO(2024-05-02 GA)\u306b\u3088\u3063\u3066 Monitor \u8a2d\u5b9a\u306a\u3057\u306b SLO \u4f5c\u6210\u4e2d\u306b\u3057\u304d\u3044\u5024\u3092\u8a66\u884c\u3067\u304d\u308b\u3088\u3046\u306b\u3057\u305f\u3002\u3055\u3089\u306b 1 \u3064\u306e SLI \u306b\u5bfe\u3057\u3066 Upside(\u5c06\u6765\u76ee\u6307\u3059\u7406\u60f3\u5024)\u30fbDownside(\u73fe\u5b9f\u7684\u306b\u5272\u308a\u305f\u304f\u306a\u3044\u73fe\u72b6\u76e3\u8996\u6c34\u6e96)\u30fbActual(\u65e2\u306b\u4e00\u90e8\u5272\u308c\u3066\u3044\u308b\u6c34\u6e96)\u306e 3 \u6bb5\u968e\u3092\u5b9a\u7fa9\u3059\u308b **Multi-tiered SLOs** \u3092\u63a1\u7528\u3057\u3001\u958b\u767a\u30d5\u30a7\u30fc\u30ba\u304c\u9032\u3080\u306b\u3064\u308c\u3066 Actual \u2192 Downside \u306e\u307f\u3092\u6ce8\u8996\u3059\u308b\u72b6\u614b\u3078\u53ce\u6582\u3055\u305b\u308b\u8a2d\u8a08\u3068\u3057\u305f\u3002\u6280\u8853\u7684\u306a\u8a08\u6e2c\u57fa\u76e4\u306e\u6574\u5099\u4ee5\u4e0a\u306b\u3001SLI \u30c0\u30c3\u30b7\u30e5\u30dc\u30fc\u30c9\u306e\u6574\u5099\u3068 Weekly \u30ec\u30d3\u30e5\u30fc\u306b\u3088\u308b\u300c\u6587\u5316\u91b8\u6210\u300d\u304c\u6700\u91cd\u8981\u3068\u4f4d\u7f6e\u3065\u3051\u3089\u308c\u3066\u3044\u308b [42]\u3002\n\n### 8.8 AI \u30b5\u30fc\u30d3\u30b9\u306e SLI/SLO\n\n\u751f\u6210 AI \u30b5\u30fc\u30d3\u30b9\u3067\u306f SLI \u306e\u5019\u88dc\u304c\u300c\u5fdc\u7b54\u306e\u54c1\u8cea\u300d\u3078\u5e83\u304c\u308b\u3002Yoshikawa\uff08Topotal\uff09\u306f\u3001\u51fa\u529b\u54c1\u8cea\u30b9\u30b3\u30a2\u3001\u30cf\u30eb\u30b7\u30cd\u30fc\u30b7\u30e7\u30f3\u7387\u3001RAG \u691c\u7d22\u7cbe\u5ea6\u3092 SLI \u306b\u52a0\u3048\u308b\u4f8b\u3092\u793a\u3057\u305f\u3002LLM \u63a8\u8ad6\u3067\u306f TTFT(Time To First Token)\u30fbITL(Inter Token Latency)\u30fbE2EL(End-to-End Latency)\u30fbGoodput(\u30ec\u30a4\u30c6\u30f3\u30b7\u8981\u4ef6\u3092\u6e80\u305f\u3059\u30ea\u30af\u30a8\u30b9\u30c8\u6570)\u306b\u52a0\u3048\u3001Tokens/Dollar \u3084 Tokens/User/Dollar \u3092\u8cbb\u7528\u5bfe\u52b9\u679c\u6307\u6a19\u3068\u3057\u3066\u4f4d\u7f6e\u3065\u3051\u308b [43][44]\u3002\n\n### 8.9 \u30ab\u30fc\u30dc\u30f3\u8a8d\u8b58 SLO\n\nCASCA(Carbon-Aware SLO and Control plAtform)\u306f\u30ab\u30fc\u30dc\u30f3\u30d5\u30c3\u30c8\u30d7\u30ea\u30f3\u30c8\u3092\u5831\u916c\u95a2\u6570\u306b\u7d44\u307f\u8fbc\u307f\u3001SLO \u5145\u8db3\u3068\u30b5\u30b9\u30c6\u30ca\u30d3\u30ea\u30c6\u30a3\u306e\u4e21\u7acb\u3092\u56f3\u308b\u30de\u30a4\u30af\u30ed\u30b5\u30fc\u30d3\u30b9\u539f\u5247\u30d9\u30fc\u30b9\u306e\u30d7\u30e9\u30c3\u30c8\u30d5\u30a9\u30fc\u30e0\u3067\u3042\u308b\u3002\u30a4\u30f3\u30d5\u30e9\u30d7\u30ed\u30d0\u30a4\u30c0\u30fc\u304c\u30b5\u30fc\u30d3\u30b9\u306e\u30bb\u30de\u30f3\u30c6\u30a3\u30af\u30b9\u3092\u77e5\u308b\u3053\u3068\u306a\u304f SLO \u5145\u8db3\u306e\u305f\u3081\u306b\u52d5\u7684\u518d\u8a2d\u5b9a\u3092\u884c\u3048\u308b\u3088\u3046\u3001SLO \u5ba3\u8a00\u3068\u30d1\u30e9\u30e1\u30fc\u30bf\u540d\u3092\u300cServiceSLO\u300d\u300cServiceParam\u300d\u3078\u533f\u540d\u5316\u3059\u308b\u4ed5\u7d44\u307f\u3092\u6301\u3064\u3002\u533f\u540d\u5316\u3057\u3066\u3082\u610f\u601d\u6c7a\u5b9a\u30b7\u30b9\u30c6\u30e0\u306e\u6319\u52d5\u306f\u307b\u307c\u5909\u5316\u305b\u305a(SLO \u5145\u8db3\u7387\u30fb\u30ab\u30fc\u30dc\u30f3\u30d5\u30c3\u30c8\u30d7\u30ea\u30f3\u30c8\u3068\u3082\u5b8c\u5168\u8a18\u8ff0\u6642\u3068\u540c\u69d8\u306e\u5e8f\u5217)\u3001\u30d7\u30e9\u30a4\u30d0\u30b7\u30fc\u4fdd\u8b77\u3068 SLO \u5145\u8db3\u3092\u4e21\u7acb\u3067\u304d\u308b\u3053\u3068\u3092\u5b9f\u8a3c\u3057\u305f\u3002\n\n\u8caa\u6b32\u30d2\u30e5\u30fc\u30ea\u30b9\u30c6\u30a3\u30c3\u30af(GDS)\u30fb\u5f37\u5316\u5b66\u7fd2(RLDS)\u30fb\u30e9\u30f3\u30c0\u30e0\u57fa\u6e96(RDS)\u306e 3 \u65b9\u5f0f\u3092 24 \u6642\u9593\u306e\u5b9f\u6a5f\u8a55\u4fa1\u3067\u6bd4\u8f03\u3057\u305f\u3068\u3053\u308d\u3001FPS SLO \u5145\u8db3\u7387\u3067\u306f GDS \u304c\u6700\u826f(90.6%\u3001RLDS: 85.2%)\u3067\u3042\u308b\u4e00\u65b9\u3001\u30ab\u30fc\u30dc\u30f3\u30d5\u30c3\u30c8\u30d7\u30ea\u30f3\u30c8\u306e\u6700\u5c0f\u5316\u3067\u306f\u5831\u916c\u95a2\u6570\u306b\u30ab\u30fc\u30dc\u30f3\u5f37\u5ea6\u3092\u7d44\u307f\u8fbc\u3093\u3060 RLDS \u304c\u6700\u826f(72.4 vs GDS 74.9 mgCO\u2082/\u5206)\u3068\u306a\u308a\u3001**\u30d1\u30d5\u30a9\u30fc\u30de\u30f3\u30b9\u6700\u9069\u5316\u3068\u30b5\u30b9\u30c6\u30ca\u30d3\u30ea\u30c6\u30a3\u6700\u9069\u5316\u306f\u30c8\u30ec\u30fc\u30c9\u30aa\u30d5\u306e\u95a2\u4fc2\u306b\u3042\u308b**\u3053\u3068\u3092\u5b9a\u91cf\u7684\u306b\u793a\u3057\u305f\u3002\u5ba3\u8a00\u7684\u306a SLO\u30fb\u30d1\u30e9\u30e1\u30fc\u30bf\u306e\u8ffd\u52a0\u5909\u66f4\u306f\u5e73\u5747 2.8 \u79d2\u3067\u5b8c\u4e86\u3057\u3001\u30bd\u30fc\u30b9\u30b3\u30fc\u30c9\u4fee\u6b63\u30fb\u518d\u30d3\u30eb\u30c9\u3092\u8981\u3059\u308b\u547d\u4ee4\u7684\u30a2\u30d7\u30ed\u30fc\u30c1(\u5e73\u5747 56.5 \u79d2)\u3068\u6bd4\u3079\u7d04 54 \u79d2\u77ed\u7e2e\u3057\u305f [45]\u3002\n\n### 8.10 \u30bb\u30eb\u30d5\u30af\u30e9\u30d5\u30c8 \u2014 SLO \u306e\u672a\u6765\u50cf\n\n\u576a\u5185\u7950\u6a39\uff08\u3055\u304f\u3089\u30a4\u30f3\u30bf\u30fc\u30cd\u30c3\u30c8\u7814\u7a76\u6240\uff09\u306f\u30012040 \u5e74\u4ee3\u306e\u500b\u5225\u5316\u30a2\u30d7\u30ea\u30b1\u30fc\u30b7\u30e7\u30f3\u3067\u306f\u4fe1\u983c\u6027\u30fb\u30b3\u30b9\u30c8\u30fb\u5909\u66f4\u901f\u5ea6\u306a\u3069\u306e\u57fa\u672c\u5909\u91cf\u306e\u5747\u8861\u70b9\u3092\u5229\u7528\u8005\u3068 AI \u304c\u5bfe\u8a71\u7684\u304b\u3064\u4f53\u9a13\u7684\u306b\u6c7a\u3081\u308b\u3068\u63cf\u3044\u305f\u3002SLO \u3092\u9759\u7684\u306a\u5951\u7d04\u5024\u3067\u306f\u306a\u304f\u3001\u63a2\u7d22\u30d7\u30ed\u30bb\u30b9\u3068\u3057\u3066\u6271\u3046\u8996\u70b9\u3067\u3042\u308a\u3001SLO \u306e\u4e3b\u4f53\u304c\u4e8b\u696d\u8005\u304b\u3089\u5229\u7528\u8005+AI \u306b\u79fb\u884c\u3059\u308b\u65b9\u5411\u6027\u304c\u793a\u5506\u3055\u308c\u308b [46]\u3002\n\n### 8.11 \u30bb\u30ad\u30e5\u30ea\u30c6\u30a3\u9818\u57df\u3078\u306e\u5fdc\u7528 \u2014 Security Level Objectives\n\nJohn Benninghoff\uff08Security Differently\uff09\u306f SREcon25 Americas \u3067\u3001SRE \u306e SLO \u3092\u30bb\u30ad\u30e5\u30ea\u30c6\u30a3\u9818\u57df\u306b\u8ee2\u7528\u3057\u305f **Security Level Objectives** \u3092\u63d0\u5531\u3057\u305f\u3002\u7740\u60f3\u306f Safety-II \u306e\u67a0\u7d44\u307f(\u60aa\u3044\u7d50\u679c\u306e\u975e\u767a\u751f\u3067\u306a\u304f\u826f\u3044\u7d50\u679c\u306e\u5897\u52a0\u3092\u76ee\u6307\u3059\u3001\u5206\u5e03\u306e\u53f3\u30b7\u30d5\u30c8)\u306b\u57fa\u3065\u304f\u30022019 \u5e74\u306e DORA\u30fbVeracode\u30fbSonatype \u3068\u3044\u3046 3 \u3064\u306e\u72ec\u7acb\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u306f\u3001\u30b5\u30a4\u30d0\u30fc\u30bb\u30ad\u30e5\u30ea\u30c6\u30a3\u306e\u30d1\u30d5\u30a9\u30fc\u30de\u30f3\u30b9\u3068\u30bd\u30d5\u30c8\u30a6\u30a7\u30a2\u30c7\u30ea\u30d0\u30ea\u30fc\u30fb\u4fe1\u983c\u6027\u30d1\u30d5\u30a9\u30fc\u30de\u30f3\u30b9\u304c\u5f37\u304f\u76f8\u95a2\u3059\u308b\u3053\u3068\u3092\u65e2\u306b\u793a\u3057\u3066\u304a\u308a\u3001\u5b9f\u8a3c\u7684\u306b\u52b9\u679c\u306e\u9ad8\u3044\u30bb\u30ad\u30e5\u30ea\u30c6\u30a3\u30b3\u30f3\u30c8\u30ed\u30fc\u30eb\u306e\u30c8\u30c3\u30d7 2(\u653b\u6483\u9762\u7ba1\u7406\u30fb\u30d1\u30c3\u30c1\u983b\u5ea6)\u306f SRE \u306e\u4e2d\u6838\u7684\u306a\u904b\u7528\u6d3b\u52d5(\u30a4\u30f3\u30d9\u30f3\u30c8\u30ea\u30fb\u69cb\u6210\u7ba1\u7406\u3001\u4f9d\u5b58\u95a2\u4fc2\u7ba1\u7406)\u3068 1 \u5bfe 1 \u3067\u5bfe\u5fdc\u3059\u308b [47]\u3002\n\nSecurity Level Objectives \u306e\u5019\u88dc\u6307\u6a19\u306b\u306f\u3001\u30a8\u30f3\u30c9\u30dd\u30a4\u30f3\u30c8\u5f53\u305f\u308a\u306e\u8106\u5f31\u6027\u6570\u3001\u516c\u958b\u30b5\u30fc\u30d3\u30b9\u5f53\u305f\u308a\u306e\u958b\u653e\u30dd\u30fc\u30c8\u6570\u3001MFA \u672a\u4f7f\u7528\u30a8\u30f3\u30c9\u30dd\u30a4\u30f3\u30c8\u7387\u3001\u5b64\u7acb\u30a2\u30ab\u30a6\u30f3\u30c8\u7387\u3001\u30ed\u30b0\u30a4\u30f3\u5931\u6557\u7387\u30fb\u6210\u529f\u7387\u304c\u6319\u3052\u3089\u308c\u308b\u3002\u901a\u5e38\u306e SLO \u3068\u306e\u672c\u8cea\u7684\u306a\u9055\u3044\u306f\u640d\u5931\u5206\u5e03\u306e\u898f\u6a21\u306b\u3042\u308b\u3002\u30a2\u30a6\u30c6\u30fc\u30b8\u640d\u5931\u304c $100\u2013$10M(\u5178\u578b $100K)\u3067\u3042\u308b\u306e\u306b\u5bfe\u3057\u3001\u30b5\u30a4\u30d0\u30fc\u30bb\u30ad\u30e5\u30ea\u30c6\u30a3\u640d\u5931\u306f $100\u2013$10B(\u5178\u578b $1M\u300195 \u30d1\u30fc\u30bb\u30f3\u30bf\u30a4\u30eb $52M)\u3068\u6841\u9055\u3044\u306b\u5927\u304d\u304f\u3001\u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u7684\u306a\u300c\u8a31\u5bb9\u3067\u304d\u308b\u640d\u5931\u7387\u300d\u306e\u8a2d\u5b9a\u304c\u56f0\u96e3\u3067\u3042\u308b\u3002\u305d\u306e\u305f\u3081 Security Level Objectives \u306f\u640d\u5931\u7387\u305d\u306e\u3082\u306e\u3067\u306f\u306a\u304f\u3001\u30ea\u30b9\u30af\u3068\u76f8\u95a2\u3059\u308b\u5148\u884c\u6307\u6a19\u306b\u95be\u5024\u3092\u8a2d\u5b9a\u3059\u308b\u8a2d\u8a08\u306b\u5bc4\u308b [47]\u3002\u3053\u306e\u69cb\u9020\u306f\u7b2c 4.6 \u7bc0\u306e\u8106\u5f31\u6027\u30d0\u30b8\u30a7\u30c3\u30c8(Thomson\u30fbLaing)\u3068\u540c\u3058\u554f\u984c\u9818\u57df\u3092\u6271\u3046\u304c\u3001Benninghoff \u306f\u300cSRE \u306e\u65e5\u5e38\u904b\u7528\u305d\u306e\u3082\u306e\u304c\u30bb\u30ad\u30e5\u30ea\u30c6\u30a3\u30d1\u30d5\u30a9\u30fc\u30de\u30f3\u30b9\u306e\u571f\u53f0\u3067\u3042\u308b\u300d\u3068\u3044\u3046\u7d44\u7e54\u8ad6\u7684\u8996\u70b9\u3092\u52a0\u3048\u3066\u3044\u308b\u70b9\u3067\u7570\u306a\u308b\u3002\n\n### 8.12 \u5b9a\u5e38\u6027\u30e2\u30c7\u30eb \u2014 \u95be\u5024\u30d9\u30fc\u30b9 SLI/SLO \u3092\u8d85\u3048\u308b\u4ee3\u66ff\u30a2\u30d7\u30ed\u30fc\u30c1\n\nNarayan Desai(Google Cloud SRE)\u306f SREcon21 \u3067\u3001SLI \u306b\u56fa\u5b9a\u95be\u5024\u3092\u8a2d\u5b9a\u3059\u308b\u5f93\u6765\u30a2\u30d7\u30ed\u30fc\u30c1\u3092\u300cGoldilocks Reliability\u300d\u3068\u547c\u3073\u3001\u3053\u308c\u304c\u6697\u9ed9\u306b\u4f9d\u5b58\u3059\u308b 4 \u3064\u306e\u8377\u91cd\u4eee\u5b9a \u2015\u2015 \u300c\u3061\u3087\u3046\u3069\u3044\u3044\u300d\u306e\u6709\u610f\u6027\u30fb\u7b54\u3048\u306e\u4e00\u610f\u6027\u30fb\u554f\u3044\u306e\u65e2\u77e5\u6027\u30fb\u7b54\u3048\u306e\u4e0d\u5909\u6027 \u2015\u2015 \u3092\u660e\u793a\u5316\u3057\u305f\u4e0a\u3067\u6279\u5224\u3057\u305f\u3002\u95be\u5024\u901a\u904e\u5224\u5b9a\u306f\u5143\u306e\u5206\u5e03\u3092 2 \u30d0\u30b1\u30c3\u30c8\u30d2\u30b9\u30c8\u30b0\u30e9\u30e0\u3078\u5727\u7e2e\u3059\u308b\u64cd\u4f5c\u3067\u3042\u308a\u3001Hartmann(2019\u3001\u7b2c 2.2 \u7bc0)\u304c\u6307\u6458\u3057\u305f\u30d1\u30fc\u30bb\u30f3\u30bf\u30a4\u30eb\u96c6\u7d04\u4e0d\u80fd\u554f\u984c\u3068\u540c\u6839\u306e\u300c\u70b9\u63a8\u5b9a\u306b\u3088\u308b\u30c7\u30fc\u30bf\u5727\u7e2e\u300d\u554f\u984c\u3092\u62b1\u3048\u308b [48]\u3002\n\n\u4ee3\u66ff\u3068\u3057\u3066\u63d0\u793a\u3055\u308c\u305f\u306e\u304c**\u5b9a\u5e38\u6027\u30e2\u30c7\u30eb(Stationarity-based Reliability Model)**\u3067\u3042\u308b\u3002\u4fe1\u983c\u6027\u3092\u53ef\u7528\u6027\u30fb\u30d1\u30d5\u30a9\u30fc\u30de\u30f3\u30b9\u30fb\u6b63\u78ba\u6027\u306e 3 \u6b21\u5143\u306b\u5206\u89e3\u3057\u3001\u305d\u308c\u305e\u308c\u306b\u5b9a\u5e38\u6027\u4eee\u5b9a(\u53ef\u7528\u6027=\u30a8\u30e9\u30fc\u306e\u6642\u7a7a\u9593\u7684 i.i.d.\u3001\u30d1\u30d5\u30a9\u30fc\u30de\u30f3\u30b9=\u9577\u6642\u9593\u7a93\u3067\u306e\u4e00\u5b9a\u6027\u3001\u6b63\u78ba\u6027=\u30d0\u30b0\u3092\u9664\u304f\u540c\u4e00\u5165\u529b\u304b\u3089\u306e\u540c\u4e00\u51fa\u529b)\u3092\u4ed8\u4e0e\u3057\u3001\u305d\u306e\u9038\u8131\u3092\u4fe1\u983c\u6027\u52a3\u5316\u306e\u30b7\u30b0\u30ca\u30eb\u3068\u3057\u3066\u691c\u77e5\u3059\u308b\u3002\u30ad\u30e3\u30ea\u30d6\u30ec\u30fc\u30b7\u30e7\u30f3\u4e0d\u8981(\u95be\u5024\u8a2d\u5b9a\u306e\u305f\u3081\u306e\u4eba\u9593\u5224\u65ad\u304c\u4e0d\u8981)\u3067\u3001\u5b9a\u5e38\u6027\u9055\u53cd\u3092\u968e\u5c64\u7684\u306b\u5206\u89e3\u3059\u308b\u3053\u3068\u3067\u6839\u672c\u539f\u56e0\u3092\u8b58\u5225\u3067\u304d\u308b(\u4f8b: \u300c\u30b9\u30ed\u30fc\u30af\u30a8\u30ea\u5272\u5408\u300d\u5168\u4f53\u306e\u5b9a\u5e38\u6027\u9055\u53cd \u2192 \u7406\u7531\u5225\u306b\u5206\u89e3\u3057 IO Time \u304c\u539f\u56e0\u3068\u7279\u5b9a)\u3002\u3053\u306e\u624b\u6cd5\u306f 2022 \u5e74\u306e SREcon22 Americas(Principled Performance Analytics)\u3067\u6b63\u898f\u5206\u5e03 z \u30b9\u30b3\u30a2\u306b\u3088\u308b 2\u03c3 \u624b\u6cd5\u3068\u3057\u3066\u6570\u7406\u5b9f\u88c5\u3055\u308c\u3001Google \u672c\u756a\u74b0\u5883\u3067\u969c\u5bb3\u306e 18 \u6642\u9593\u5148\u884c\u691c\u77e5\u3092\u5b9f\u8a3c\u3057\u305f [48][49]\u3002\n\n\u5b9a\u5e38\u6027\u30e2\u30c7\u30eb\u306f SLI/SLO \u306e\u95be\u5024\u8a2d\u8a08\u554f\u984c\u305d\u306e\u3082\u306e\u3092\u89e3\u6d88\u3059\u308b\u306e\u3067\u306f\u306a\u304f\u3001SLO \u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af\u3068**\u76f4\u4ea4**\u3059\u308b\u300c\u3042\u308b\u3079\u304d\u72b6\u614b\u304b\u3089\u306e\u9038\u8131\u300d\u3092\u691c\u77e5\u3059\u308b\u5c64\u3068\u3057\u3066\u4f4d\u7f6e\u3065\u3051\u3089\u308c\u308b\u3002SLO \u304c\u300c\u3069\u3053\u3092\u8d85\u3048\u305f\u3089\u30a2\u30a6\u30c8\u304b\u300d\u3092\u5b9a\u7fa9\u3059\u308b\u306e\u306b\u5bfe\u3057\u3001\u5b9a\u5e38\u6027\u30e2\u30c7\u30eb\u306f\u300c\u901a\u5e38\u3069\u3046\u3042\u308b\u3079\u304d\u304b\u300d\u3092\u5b9a\u7fa9\u3059\u308b\u70b9\u304c\u6839\u672c\u7684\u306b\u7570\u306a\u308b\u3002\n\n---\n\n## \u7b2c 9 \u7ae0\u3000\u672a\u89e3\u6c7a\u306e\u554f\u3044\u3068\u30d5\u30ed\u30f3\u30c6\u30a3\u30a2\n\n### 9.1 SLO \u4ee3\u6570\n\n\u8907\u6570\u306e\u30c0\u30a6\u30f3\u30b9\u30c8\u30ea\u30fc\u30e0\u30b5\u30fc\u30d3\u30b9\u3092\u6d88\u8cbb\u3059\u308b\u4e0a\u4f4d\u30b5\u30fc\u30d3\u30b9\u306e SLO \u3092\u3001\u4f9d\u5b58\u5148\u306e SLO \u304b\u3089\u5206\u6790\u7684\u306b\u5c0e\u51fa\u3059\u308b\u4f53\u7cfb\u7684\u624b\u6cd5\u304c\u5b58\u5728\u3057\u306a\u3044\u3002\u4f9d\u5b58\u95a2\u4fc2\u306e\u76f4\u5217/\u4e26\u5217\u6d88\u8cbb\u30d1\u30bf\u30fc\u30f3\u30fb\u30d5\u30a7\u30a4\u30eb\u30aa\u30fc\u30d7\u30f3/\u30af\u30ed\u30fc\u30ba\u30c9\u52d5\u4f5c\u3092\u8003\u616e\u3057\u305f\u4ee3\u6570\u7684\u67a0\u7d44\u307f\u306e\u78ba\u7acb\u306f\u672a\u89e3\u6c7a\u8ab2\u984c [15]\u3002\n\n### 9.2 SLO \u306e\u6839\u672c\u6279\u5224\n\nDesai \u306f\u300cSLO requires recognized errors\u300d\u3068\u8ff0\u3079\u305f\u4e0a\u3067\u3001\u30a8\u30e9\u30fc\u8a8d\u8b58\u304c\u66d6\u6627\u30fb\u30d0\u30b0\u7531\u6765\u30fb\u8f03\u6b63\u8aa4\u5dee\u30fb\u5b9a\u671f\u30e1\u30f3\u30c6\u306a\u3057\u3068\u3044\u3046 4 \u3064\u306e\u69cb\u9020\u7684\u554f\u984c\u304b\u3089\u300c\u30a8\u30e9\u30fc\u306f\u6d45\u3044\u30c7\u30fc\u30bf\u306b\u3068\u3069\u307e\u308b\u300d\u3068\u4e3b\u5f35\u3057\u305f\u3002SLO \u3092\u6bb5\u968e\u7684\u6539\u5584\u306e\u5bfe\u8c61\u3068\u3057\u3066\u8ad6\u3058\u308b\u591a\u304f\u306e\u5148\u884c\u30bd\u30fc\u30b9\u3068\u306f\u7570\u306a\u308a\u3001\u300cSLO \u305d\u306e\u3082\u306e\u304c\u8a08\u6e2c\u30c4\u30fc\u30eb\u3068\u3057\u3066\u6839\u672c\u7684\u306b\u4e0d\u5b8c\u5168\u300d\u3068\u3044\u3046\u7d50\u8ad6\u306b\u9032\u3093\u3060\u6700\u3082\u5f37\u3044\u6279\u5224\u5f62\u3067\u3042\u308b [49]\u3002\n\n### 9.3 \u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u306e\u4e0d\u78ba\u5b9f\u6027\n\n\u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u3092\u8a2d\u5b9a\u3059\u308b\u305f\u3081\u306e\u30a4\u30f3\u30d1\u30af\u30c8\u63a8\u5b9a\u304c\u6841\u9055\u3044\u306b\u4e0d\u6b63\u78ba\u3067\u3042\u308a\u3046\u308b\u3002Davidovi\u010d\uff08Google\uff09\u306f 3 \u4eba\u306e\u72ec\u7acb\u3057\u305f\u62c5\u5f53\u8005\u306b\u30a4\u30f3\u30b7\u30c7\u30f3\u30c8\u5f71\u97ff\u3092\u63a8\u5b9a\u3055\u305b\u3066\u30d0\u30ea\u30a2\u30f3\u30b9\u3092\u89b3\u5bdf\u3059\u308b\u3053\u3068\u3092\u63a8\u5968\u3057\u305f [26]\u3002\n\n### 9.4 \u305d\u306e\u4ed6\u306e\u672a\u89e3\u6c7a\u8ab2\u984c\n\n- SLE/CBE \u67a0\u7d44\u307f\u306f\u6982\u5ff5\u7684\u63d0\u6848\u306b\u3068\u3069\u307e\u308a\u3001\u30af\u30e9\u30a6\u30c9\u30d7\u30ed\u30d0\u30a4\u30c0\u3067\u306e\u63a1\u7528\u4e8b\u4f8b\u306f\u672a\u5831\u544a\n- AI \u30b5\u30fc\u30d3\u30b9\u3067\u30cf\u30eb\u30b7\u30cd\u30fc\u30b7\u30e7\u30f3\u7387\u3092\u3069\u306e\u30e6\u30fc\u30b6\u30fc\u5e78\u798f\u30d5\u30ec\u30fc\u30d0\u30fc\u306b\u5206\u985e\u3059\u308b\u304b\u3001\u307e\u305f\u306f 7 \u756a\u76ee\u306e\u30d5\u30ec\u30fc\u30d0\u30fc\u304c\u5fc5\u8981\u304b\n- SLO \u30a2\u30e9\u30fc\u30c8\u767a\u706b\u6642\u306b\u81ea\u52d5\u6dfb\u4ed8\u3059\u308b\u60c5\u5831\u306e\u6a19\u6e96\u5316\u7bc4\u56f2\n- \u30af\u30e9\u30a4\u30a2\u30f3\u30c8\u5074\u6e2c\u5b9a\u3068\u30b5\u30fc\u30d0\u30fc\u5074\u6e2c\u5b9a\u306e\u4e56\u96e2\u3092\u5b9a\u91cf\u5316\u3059\u308b\u5b9f\u8df5\u7684\u624b\u9806\n\n---\n\n## \u4ed8\u9332\n\n### A. \u6587\u732e\u5e74\u8868\n\n| \u5e74 | \u6587\u732e | \u4e3b\u306a\u8ca2\u732e |\n|---|---|---|\n| 1984 | \u9ad8\u901f\u9053\u8def Levels of Service | SLI/SLO \u306e\u6982\u5ff5\u7684\u8d77\u6e90 |\n| 2007 | Hamilton, LISA | SLA \u3092\u8a2d\u8a08\u6bb5\u968e\u306e\u9178\u6027\u8a66\u9a13\u3068\u4f4d\u7f6e\u3065\u3051 |\n| 2014 | Treynor Sloss, SREcon14 | 13 \u306e\u30ad\u30fc\u3068\u300c\u30ed\u30fc\u30f3\u30c1\u30aa\u30f3\u30d6\u30e9\u30c3\u30af\u300d\u30eb\u30fc\u30eb\u521d\u51fa |\n| 2015 | Alvidrez, SREcon15 | \u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u306e\u8d77\u6e90\u4f53\u9a13 |\n| 2016 | SRE Book Ch3-4 | SLI/SLO/SLA \u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af\u4f53\u7cfb\u5316 |\n| 2016 | Jones+Murphy, SREcon16 | SLO \u516c\u8868\u7406\u7531\u3001\u96c6\u7d04\u30a4\u30f3\u30bf\u30fc\u30d0\u30eb\u554f\u984c\u3001\u30d0\u30f3\u30d0\u30f3\u5236\u5fa1\u56de\u907f |\n| 2017 | Mogul+, HotOS | \u53ef\u7528\u6027\u5b9a\u7fa9\u306e\u56f0\u96e3\u3055\u3092\u63d0\u8d77 |\n| 2017 | Wilkinson, SREcon17 | Prometheus \u3067\u306e\u30d0\u30b1\u30c3\u30c8\u6bd4\u7387 SLI \u5b9f\u88c5 |\n| 2018 | SRE Workbook | SLO \u5b9f\u88c5\u30fb\u30a2\u30e9\u30fc\u30c6\u30a3\u30f3\u30b0\u30fb\u4e8b\u4f8b\u30fb\u6587\u66f8\u4f8b\u3092\u4f53\u7cfb\u5316 |\n| 2018 | Wilkinson, SREcon18 Asia | SLO \u30d0\u30fc\u30f3\u30ec\u30fc\u30c8\u30a2\u30e9\u30fc\u30c8\u3068\u30b7\u30f3\u30d7\u30c8\u30e0\u30d9\u30fc\u30b9\u30c9\u30a2\u30e9\u30fc\u30c6\u30a3\u30f3\u30b0 |\n| 2018 | Binette+Flaming, SREcon18 Americas | \u30b1\u30a4\u30d1\u30d3\u30ea\u30c6\u30a3\u99c6\u52d5 SLI \u8a2d\u8a08 |\n| 2018 | Gangatirkar, SREcon18 Asia | \u5171\u611f\u30ae\u30e3\u30c3\u30d7\u3001S \u5b57\u66f2\u7dda\u30016 \u30d5\u30ec\u30fc\u30d0\u30fc |\n| 2018 | Vieiro, SREcon18 Asia | Atlassian Error Budgets 0.1 |\n| 2019 | Mogul+Wilkes, HotOS | SLE/CBE\u3001SLOgician |\n| 2019 | Hartmann/Moyer, SREcon19 | \u30ec\u30a4\u30c6\u30f3\u30b7 SLO \u306e\u30d1\u30fc\u30bb\u30f3\u30bf\u30a4\u30eb\u96c6\u7d04\u4e0d\u80fd\u554f\u984c |\n| 2019 | Desai, SREcon19 EMEA | SLO \u306e 4 \u30e6\u30fc\u30b9\u30b1\u30fc\u30b9\u3001\u6697\u9ed9\u7684\u4eee\u5b9a\u3001SLO Algebra |\n| 2019 | Booking.com, SREcon19 EMEA | \u30c7\u30fc\u30bf\u54c1\u8cea SLO |\n| 2019 | Lawson, SREcon19 Americas | \u65b0\u898f\u30b5\u30fc\u30d3\u30b9\u3078\u306e SLO \u5b9f\u88c5 6 \u30b9\u30c6\u30c3\u30d7 |\n| 2020 | Hauer+, NSDI | \u30a6\u30a3\u30f3\u30c9\u30a6\u4ed8\u304d\u30e6\u30fc\u30b6\u30fc\u30a2\u30c3\u30d7\u30bf\u30a4\u30e0 |\n| 2020 | Coulter, SREcon20 Americas | SLO \u4ea4\u6e09\u30d5\u30ed\u30fc\u3001\u30e6\u30fc\u30b6\u30fc\u884c\u52d5\u30d9\u30fc\u30b9 SLI |\n| 2020 | Moyer, SREcon20 Americas | \u30a8\u30e9\u30fc\u30d0\u30b8\u30a7\u30c3\u30c8\u6c11\u4e3b\u5316\u306e 3 \u9375 |\n| 2021 | Ding+Zhang, SREcon21 | SLX \u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af |\n| 2021 | Desai, SREcon21 | Goldilocks Reliability \u6279\u5224\u3001\u5b9a\u5e38\u6027\u30e2\u30c7\u30eb |\n| 2022 | Palcuie, SREcon22 EMEA | 30\u21923000 \u4e07 SLO\u3001\u9867\u5ba2\u5358\u4f4d SLO |\n| 2022 | Davidovi\u010d, SREcon22 EMEA | SLO \u30e2\u30c7\u30eb\u306e\u7dda\u5f62\u6027\u554f\u984c |\n| 2022 | Desai, SREcon22 Americas | SLO \u6839\u672c\u6279\u5224 |\n| 2022 | \u8fd1\u85e4, SRE NEXT 2022 | \u975e\u6a5f\u80fd\u8981\u6c42\u4e88\u7b97\u306e\u7d44\u7e54\u7684\u524d\u63d0 |\n| 2023 | Koss+Goins, SREcon23 Americas | \u6642\u9593\u30b9\u30e9\u30a4\u30b9 vs \u30a4\u30d9\u30f3\u30c8\u30d9\u30fc\u30b9\u96c6\u8a08 |\n| 2023 | Furino, SREcon23 EMEA | \u6210\u529f\u5b9a\u7fa9 SLI\u30013 \u30da\u30eb\u30bd\u30ca\u6642\u9593\u7a93 |\n| 2024 | Sedlak+, SOSE | SLO \u62e1\u6563\u65b9\u6cd5\u8ad6 |\n| 2024 | Tsuda, SRE NEXT 2024 | \u30af\u30e9\u30a4\u30a2\u30f3\u30c8\u30b5\u30a4\u30c9 SLO\u3001Multi-tiered SLOs |\n| 2025 | Durst, SREcon25 EMEA | SLO \u5c0e\u5165\u306e 4 \u524d\u63d0\u6761\u4ef6 |\n| 2025 | Riot Games, SREcon25 Americas | Player Minutes \u3068 CEO OKR |\n| 2025 | eBay, SREcon25 Americas | \u975e\u540c\u671f\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3 SLO |\n| 2025 | Benninghoff, SREcon25 Americas | Security Level Objectives |\n| 2026 | Takamura, Road to SRE NEXT 2026 | SLI/SLO \u6bb5\u968e\u7684\u5c0e\u5165\u306e\u4f53\u7cfb\u5316 |\n| 2026 | Yoshikawa | AI \u30b5\u30fc\u30d3\u30b9 SLI/SLO |\n| 2026 | CASCA | \u30ab\u30fc\u30dc\u30f3\u8a8d\u8b58 SLO |\n\n### B. \u53c2\u8003\u6587\u732e\n\n[1] Last9. \"The Origin of Service Level Objectives.\" Last9 Blog.\nhttps://blog.last9.io/the-origin-of-service-level-objectives/\n\n[2] Ben Treynor Sloss. \"Keys to SRE.\" SREcon14, USENIX, 2014.\nhttps://www.usenix.org/conference/srecon14/technical-sessions/presentation/keys-sre\n\n[3] Chris Jones, John Wilkes, Niall Murphy, Cody Smith. \"Site Reliability Engineering - Chapter 4: Service Level Objectives.\" *Site Reliability Engineering*, O'Reilly, 2016.\nhttps://sre.google/sre-book/service-level-objectives/\n\n[4] Steven Thurgood, David Ferguson, Alex Hidalgo, Betsy Beyer. \"The Site Reliability Workbook - Chapter 2: Implementing SLOs.\" *The Site Reliability Workbook*, O'Reilly, 2018.\nhttps://sre.google/workbook/implementing-slos/\n\n[5] Jeffrey C. Mogul \u307b\u304b. \"Nines are Not Enough: Meaningful Metrics for Clouds.\" HotOS '19, ACM, 2019.\nhttps://dl.acm.org/doi/10.1145/3317550.3321432\n\n[6] Marc Alvidrez. \"Site Reliability Engineering - Chapter 3: Embracing Risk.\" *Site Reliability Engineering*, O'Reilly, 2016.\nhttps://sre.google/sre-book/embracing-risk/\n\n[7] Heinrich Hartmann. \"Latency SLOs Done Right.\" SREcon19 EMEA, USENIX, 2019.\nhttps://www.usenix.org/conference/srecon19emea/presentation/hartmann-latency\n\n[8] Chris Jones, Niall Murphy. \"Service Levels and Error Budgets.\" SREcon16, USENIX, 2016.\nhttps://www.usenix.org/conference/srecon16/program/presentation/jones\n\n[9] Elisa Binette, Matthew Flaming. \"SLOs and SLIs in the Real World: A Deep Dive.\" SREcon18 Americas, USENIX, 2018.\nhttps://www.usenix.org/conference/srecon18americas/presentation/flaming\n\n[10] Sal Furino. \"9 Things You Should Do When Starting to Use SLOs.\" SREcon23 EMEA, USENIX, 2023.\nhttps://www.usenix.org/conference/srecon23emea/presentation/furino\n\n[11] Ketan Gangatirkar. \"Quantifying Empathy Through Service Level Objectives.\" SREcon18 Asia, USENIX, 2018.\nhttps://www.usenix.org/conference/srecon18asia/presentation/gangatirkar\n\n[12] Ben McCormack, William Bonnell, Garrett Plasky, Alex Hidalgo, Betsy Beyer, Dave Rensin. \"SLO Engineering Case Studies.\" *The Site Reliability Workbook*, O'Reilly, 2018.\nhttps://sre.google/workbook/slo-engineering-case-studies/\n\n[13] Dave Stanke. \"Squish Level Objectives: How SRE can Help Align Technical Work to User Benefit.\" SREcon20 Americas, USENIX, 2020.\nhttps://www.usenix.org/conference/srecon20americas/presentation/stanke\n\n[14] Marco Coulter. \"Avoiding Goodhart's Law: Use SLO's as Tools Not Cudgels.\" SREcon20 Americas, USENIX, 2020.\nhttps://www.usenix.org/conference/srecon20americas/presentation/coulter\n\n[15] Narayan Desai. \"The Map Is Not the Territory: How SLOs Lead Us Astray, and What We Can Do about It.\" SREcon19 EMEA, USENIX, 2019.\nhttps://www.youtube.com/watch?v=NW4OOpQ3nz8\n\n[16] Steven Thurgood. \"Google SRE Workbook - Appendix A: Example SLO Document.\" *The Site Reliability Workbook*, O'Reilly, 2018.\nhttps://sre.google/workbook/slo-document/\n\n[17] Marc Alvidrez. \"Error Budgets and Risks.\" SREcon15, USENIX, 2015.\nhttps://www.usenix.org/conference/srecon15/program/presentation/alvidrez\n\n[18] Steven Thurgood. \"Google SRE Workbook - Appendix B: Example Error Budget Policy.\" *The Site Reliability Workbook*, O'Reilly, 2018.\nhttps://sre.google/workbook/error-budget-policy/\n\n[19] Troy Koss, Michael Goins. \"Not All Minutes Are Equal: The Secret behind SLO Adoption Failure.\" SREcon23 Americas, USENIX, 2023.\nhttps://www.usenix.org/conference/srecon23americas/presentation/goins\n\n[20] Jim Thomson, David Laing. \"Extending the Error Budget Model to Security and Feature Freshness.\" SREcon19 Americas, USENIX, 2019.\nhttps://www.usenix.org/conference/srecon19americas/presentation/thomson\n\n[21] Steven Thurgood, Jess Frame, Anthony Lenton, Carmela Quinito, Anton Tolchanov, Nejc Trdin. \"Alerting on SLOs.\" *The Site Reliability Workbook*, O'Reilly, 2018.\nhttps://sre.google/workbook/alerting-on-slos/\n\n[22] Alex Jones, Alois Reitbauer, Bart\u0142omiej P\u0142otka, Liz Fong-Jones, Michael Hausenblas \u307b\u304b 35 \u540d. \"Observability Whitepaper.\" CNCF TAG Observability, 2023.\nhttps://github.com/cncf/tag-observability/blob/main/whitepaper.md\n\n[23] Jamie Wilkinson. \"A Theory and Practice of Alerting with Service Level Objectives.\" SREcon18 Asia, USENIX, 2018.\nhttps://www.usenix.org/conference/srecon18asia/presentation/wilkinson\n\n[24] \u6c60\u7530\u5c06\u58eb. \"Warning\u30a2\u30e9\u30fc\u30c8\u3092\u653e\u7f6e\u3057\u306a\u3044!\u30a2\u30e9\u30fc\u30c8\u99c6\u52d5\u3067\u30ed\u30b0\u3084\u30e1\u30c8\u30ea\u30c3\u30af\u3092\u81ea\u52d5\u53ce\u96c6\u3059\u308b\u4ed5\u7d44\u307f\u306b\u3088\u308b\u6069\u6075.\" SRE NEXT 2023, 2023.\nhttps://speakerdeck.com/mashiike/warningaratowofang-zhi-sinai-aratoqu-dong-deroguyametoritukuwozi-dong-shou-ji-surushi-zu-miniyoruen-hui\n\n[25] Tam\u00e1s Hauer \u307b\u304b. \"Meaningful Availability.\" NSDI '20, USENIX, 2020.\nhttps://www.usenix.org/conference/nsdi20/presentation/hauer\n\n[26] \u0160t\u011bp\u00e1n Davidovi\u010d. \"Measuring Reliability: What Got Us Here Won't Get Us There.\" SREcon22 EMEA, USENIX, 2022.\nhttps://www.usenix.org/conference/srecon22emea/presentation/davidovic\n\n[27] Alex Palcuie. \"Going from 30 to 30 Million SLOs.\" SREcon22 EMEA, USENIX, 2022.\nhttps://www.usenix.org/conference/srecon22emea/presentation/palcuie\n\n[28] Rob Durst. \"Run, Walk, Crawl, or How We Failed Our Way to SLO Readiness.\" SREcon25 EMEA, USENIX, 2025.\nhttps://www.usenix.org/conference/srecon25emea/presentation/durst\n\n[29] Narimichi Takamura. \"\u5c0f\u3055\u304f\u306f\u3058\u3081\u308bSLI/SLO ~\u80b2\u3066\u306a\u304c\u3089\u7d44\u7e54\u306b\u5b9a\u7740\u3055\u305b\u308b\u5b9f\u8df5\u77e5~.\" Road to SRE NEXT 2026 \u795e\u6238, 2026.\nhttps://speakerdeck.com/nari_ex/slos-building-adoption-through-continuous-growth\n\n[30] \u6e21\u8fba\u8d77. \"\u30d7\u30ed\u30c0\u30af\u30c8\u30aa\u30fc\u30ca\u30fc\u3068\u3057\u3066SLO\u306b\u5411\u304d\u5408\u3046 \u301cMackerel\u30c1\u30fc\u30e0\u306e\u4e8b\u4f8b\u301c.\" SRE NEXT 2023, 2023.\nhttps://speakerdeck.com/tatsuru/sre-next-2023\n\n[31] Kristin Smith. \"Dark Sky Camping: Reducing Alert Pollution with Modern Observability Practices.\" SREcon22 Americas, USENIX, 2022.\nhttps://www.usenix.org/conference/srecon22americas/presentation/smith\n\n[32] \u8fd1\u85e4\u6b66\u58eb. \"Who owns the Service Level?\" SRE NEXT 2022, 2022.\nhttps://speakerdeck.com/chaspy/who-owns-the-service-level\n\n[33] Gui Vieiro. \"How Atlassian Is Tackling Error Budgets, Agile Style.\" SREcon18 Asia, USENIX, 2018.\nhttps://www.usenix.org/conference/srecon18asia/presentation/vieiro\n\n[34] Arnaud Lawson. \"Case Study: Implementing SLOs for a New Service.\" SREcon19 Americas, USENIX, 2019.\nhttps://www.usenix.org/conference/srecon19americas/presentation/lawson\n\n[35] Maxfield Stewart. \"Measuring Availability the Player Focused Way: How Riot Games Changed Its Availability Culture.\" SREcon25 Americas, USENIX, 2025.\nhttps://www.usenix.org/conference/srecon25americas/presentation/stewart\n\n[36] Qian Ding, Xuan Zhang. \"SLX: An Extended SLO Framework to Expedite Incident Recovery.\" SREcon21, USENIX, 2021.\nhttps://www.usenix.org/conference/srecon21/presentation/ding\n\n[37] Yoann Fouquet. \"SLOs for Data-Intensive Services.\" SREcon19 EMEA, USENIX, 2019.\nhttps://www.usenix.org/conference/srecon19emea/presentation/fouquet\n\n[38] Jash Mistry, Gabriela Medvetska. \"Beyond Sequential: A Recipe for Async Pipeline Observability and Alerting.\" SREcon25 Americas, USENIX, 2025.\nhttps://www.usenix.org/conference/srecon25americas/presentation/mistry\n\n[39] Boris Sedlak \u307b\u304b. \"Diffusing High-level SLO in Microservice Pipelines.\" SOSE 2024, IEEE, 2024.\nhttps://ieeexplore.ieee.org/document/10685329\n\n[40] Cory Lueninghoener. \"HPC Downtime Budgets: Moving SRE Practice to the Rest of the World.\" SREcon16 Europe, USENIX, 2016.\nhttps://www.usenix.org/conference/srecon16europe/program/presentation/lueninghoener\n\n[41] Wataru Tsuda. \"\u96fb\u52d5\u30de\u30a4\u30af\u30ed\u30e2\u30d3\u30ea\u30c6\u30a3\u306e\u30b7\u30a7\u30a2\u30b5\u30fc\u30d3\u30b9\u300cLUUP\u300d\u306b\u304a\u3051\u308bEnabling SLO\u306e\u5b9f\u8df5.\" SRE NEXT 2023, 2023.\nhttps://speakerdeck.com/grimoh/dian-dong-maikuromobiriteinosieasabisu-luup-niokeruenabling-slonoshi-jian\n\n[42] Wataru Tsuda. \"Enabling Client-side SLO.\" SRE NEXT 2024, 2024.\nhttps://speakerdeck.com/grimoh/enabling-client-side-slo\n\n[43] Ryota Yoshikawa. \"Reliability in the Age of AI: Engineering for AI Velocity.\" SpeakerDeck, 2026.\nhttps://speakerdeck.com/rrreeeyyy/reliability-in-the-age-of-ai-engineering-for-ai-velocity\n\n[44] \u9053\u4e0b\u5e79\u4e5f. \"\u63a8\u8ad6\u57fa\u76e4\u306e\u30d1\u30d5\u30a9\u30fc\u30de\u30f3\u30b9\u691c\u8a3c\u3068\u6700\u9069\u5316\u6226\u7565.\" SpeakerDeck, 2026.\nhttps://speakerdeck.com/aztecher/tui-lun-ji-pan-nopahuomansujian-zheng-tozui-shi-hua-zhan-lue\n\n[45] Juan Luis Herrera \u307b\u304b. \"A Microservice-Based Platform for Sustainable and Intelligent SLO Fulfilment and Service Management.\" arXiv, 2026.\nhttps://arxiv.org/abs/2602.12875\n\n[46] Yuuki Tsubouchi, Hirofumi Tsuruta. \"AI\u6642\u4ee3\u306b\u5411\u3051\u305f\u30af\u30e9\u30a6\u30c9\u306b\u304a\u3051\u308b\u4fe1\u983c\u6027\u30a8\u30f3\u30b8\u30cb\u30a2\u30ea\u30f3\u30b0\u306e\u672a\u6765\u69cb\u60f3.\" DICOMO 2022, \u60c5\u5831\u51e6\u7406\u5b66\u4f1a, 2022.\nhttps://speakerdeck.com/yuukit/dicomo2022-6a-1\n\n[47] John Benninghoff. \"Is the S in SRE for \\\"Security\\\"?\" SREcon25 Americas, USENIX, 2025.\nhttps://www.usenix.org/conference/srecon25americas/presentation/benninghoff\n\n[48] Narayan Desai. \"Beyond Goldilocks Reliability.\" SREcon21, USENIX, 2021.\nhttps://www.usenix.org/conference/srecon21/presentation/desai\n\n[49] Narayan Desai, Brent Bryan. \"Principled Performance Analytics.\" SREcon22 Americas, USENIX, 2022.\nhttps://www.usenix.org/conference/srecon22americas/presentation/desai\n", "creation_timestamp": "2026-07-06T07:24:34.707210Z"}