知识卡片
把排队与处理时间拆开
总延迟高只能说明用户等得久;排队时间高可能提示容量或调度瓶颈,处理时间高则可能提示计算、I/O或依赖变慢。分别记录接收至开始、开始至完成两段,可缩小调查范围。指标只是线索,增加实例未必能解决共享依赖瓶颈,仍要结合追踪判断。
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
- Chapter 14. Monitoring and Observability Patterns · 查看原书依据
the amount of time a request has spent in a queue, versus how much time was spent processing a request. This sort of information is especially useful when you are debugging why your service is slow in processing requests. Simply seeing that latency is high will show you
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:原文三种延迟直方图解释归因对扩容或优化决策的帮助。
- 原始内容:the amount of time a request has spent in a queue, versus how much time was spent processing a request. This sort of information is especially useful when you are debugging why your service is slow in processing requests. Simply seeing that latency is high will show you