知识卡片
性能改造先量测再假设
性能问题可能来自跨区域调用等架构路径。先测系统当前延迟与吞吐,再提出瓶颈假设并验证,避免只优化局部组件;把测量放进流水线,才能在新改动进入生产前发现退化。
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
- · 查看原书依据
Performance issues can be architectural; for example, do you have a service calling a service located in a different region or across the internet? When it comes to performance issues, measuring and creating an objective plan is critical. Take a measurement of the existing system, create a hypothesis on where performance could be improved, and then test and verify. It is important to consider the
参考来源
- 位置:《Mastering API Architecture》第8章《Redesigning Applications to API-Driven Architectures》之《Performance Issues》(源文件:_epub-src/OEBPS/ch08.xhtml)
- 结论依据:原文要求对全系统做基线测量、提出假设并测试验证,并建议自动化性能测量。
- 原始内容:Performance issues can be architectural; for example, do you have a service calling a service located in a different region or across the internet? When it comes to performance issues, measuring and creating an objective plan is critical. Take a measurement of the existing system, create a hypothesis on where performance could be improved, and then test and verify. It is important to consider the