知识卡片
删维度必须合并数值
丢弃可以删除整项指标,也可以去掉无用维度并按剩余维度聚合。书中三个pod各报100次请求,删去pod_name后应合为同一组300次,而非随意保留一个100。这样降低基数且保留总量;但失去pod级定位能力,必须先确认用途允许。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
已有反馈 · 0 条
还没有人提交反馈。
在整书笔记中的位置
原书出处 · 1 处
- Strategies for Controlling Observability Data Growth and Complexity · 查看原书依据
Dropping can mean dropping entire metrics, but it can also mean keeping only dimensions of your observability data that are useful to you and aggregating the rest in place. More often than not, even if a metric itself is very useful it will still contain dimensions that are not useful.
参考来源
- 位置:《Cloud Native Observability》第03章《Strategies for Controlling Observability Data Growth and Complexity》(来源项目 a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Cloud Native Observability/_epub-src/ch03.html#dropping_idpHQyZo)
- 结论依据:原文三个pod各100的请求数删除pod_name后汇总为300,直接显示删维度须保留剩余分组的总量。
- 原始内容:Dropping can mean dropping entire metrics, but it can also mean keeping only dimensions of your observability data that are useful to you and aggregating the rest in place. More often than not, even if a metric itself is very useful it will still contain dimensions that are not useful.