知识卡片
可观测数据分层保留
跨副本集中查询帮助定位共同错误;保存太多数据则带来成本和检索困难。降采样减少频率,聚合保留统计摘要,冷热分层保留原数据却牺牲访问速度。每种方案损失不同:均值会丢失瞬时峰值,随机丢日志会断上下文,选择前应明确未来排查和留存要求。
这张卡片看不懂?
这张卡片的反馈已满 10 条,暂时无法继续提交。
已有反馈 · 0 / 10 条
还没有人提交反馈。
在整书笔记中的位置
原书出处 · 1 处
- Chapter 14. Monitoring and Observability Patterns · 查看原书依据
This can be referred to as downsampling or aggregation. The simplest form of downsampling for metrics is to just store the metric less often. For example, if you are recording the metric every second, after a week you might reduce that to once per minute. Even this simple
参考来源
- 位置:《Designing Distributed Systems, 2nd Edition》第14章《Chapter 14. Monitoring and Observability Patterns》(来源项目 book-sources;源文件:extracted/Designing Distributed Systems, 2nd Edition/_epub-src/OEBPS/ch14.html)
- 结论依据:原文比较低频、均值、日志选择和冷存储的成本与信息损失。
- 原始内容:This can be referred to as downsampling or aggregation. The simplest form of downsampling for metrics is to just store the metric less often. For example, if you are recording the metric every second, after a week you might reduce that to once per minute. Even this simple