{
  "说明": "全部为合成输入与拟定制工作表，未连接真实系统，预期不是实测结果。",
  "研究日期": "2026-10-24",
  "题目": "压测报告只有平均响应，Agent 怎样整理能比较的性能测试证据",
  "工作表字段": [
    "实验轮次",
    "场景",
    "负载",
    "测量起止",
    "样本集合",
    "单位",
    "纳入失败口径",
    "统计方法",
    "对比限制"
  ],
  "samples": [
    {
      "case_id": "W066-01",
      "input": "A轮10用户均值200ms，B轮100用户均值250ms",
      "expected": "标负载不同，不能直接认定版本退步",
      "actual": "",
      "human_review": "由本场景授权业务或技术负责人核对；本表不自动给出批准、上线或安全通过结论。",
      "reviewer": "",
      "evidence_location": ""
    },
    {
      "case_id": "W066-02",
      "input": "两轮同场景分别给出请求时长样本100/100/900与100/100/100",
      "expected": "提供可复算样本和统计定义，实际复算栏留空",
      "actual": "",
      "human_review": "由本场景授权业务或技术负责人核对；本表不自动给出批准、上线或安全通过结论。",
      "reviewer": "",
      "evidence_location": ""
    },
    {
      "case_id": "W066-03",
      "input": "失败请求被B轮汇总剔除，A轮全部计入",
      "expected": "列统计样本口径差异，要求重出同口径材料",
      "actual": "",
      "human_review": "由本场景授权业务或技术负责人核对；本表不自动给出批准、上线或安全通过结论。",
      "reviewer": "",
      "evidence_location": ""
    }
  ],
  "补充业务输入": {
    "时长单位": "毫秒",
    "A": [
      100,
      100,
      900
    ],
    "B": [
      100,
      100,
      100
    ],
    "预期": {
      "A均值": "1100/3≈366.667",
      "B均值": 100,
      "A中位数": 100,
      "B中位数": 100,
      "说明": "合成数组的纸面复算，非压测；高分位算法未固定时不出数"
    },
    "actual": ""
  },
  "source_support": [
    {
      "url": "https://grafana.com/docs/k6/latest/using-k6/metrics/",
      "support": "Counter、Gauge、Rate、Trend用途不同；Trend可计算均值与分位数，HTTP请求、失败、时长是不同指标。",
      "opened_on": "2026-10-24"
    },
    {
      "url": "https://grafana.com/docs/k6/latest/using-k6/scenarios/",
      "support": "场景可分别设定执行器、负载与标签，开放到达率与虚拟用户等负载口径不能混为一谈。",
      "opened_on": "2026-10-24"
    }
  ]
}