知识卡片
函数架构把链路显式拆开
将键值服务拆成GET、PUT、DELETE函数后,网关负责把端点送往对应函数,函数再访问持久数据。组件虽小,失败去向、重试、观测和状态归属必须在链路层设计;局部函数单测不能替代集成行为验证。临时缓存可作为可丢弃优化,但正确性不能依赖某个执行环境持续存在。
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
- Scalability · 查看原书依据
The API gateway is configured so that requests to each of its three endpoints are directed to a different handler function — getKey, putKey, and deleteKey, respectively — which performs all of the logic for handling that request and interacting with the backing database. Granted, this is an incredibly simple application and doesn’t account for
参考来源
- 位置:《Cloud Native Go》第07章《Scalability》(来源项目book-sources;源文件:extracted/Cloud Native Go (Matthew A. Titmus)/_epub-src/OEBPS/ch07.xhtml)
- 结论依据:将键值服务拆成GET、PUT、DELETE函数后,网关负责把端点送往对应函数,函数再访问持久数据。原文相应定义、示例或限制为此结论提供依据;补充纠错与版本边界在正文就近标明。
- 原始内容:The API gateway is configured so that requests to each of its three endpoints are directed to a different handler function — getKey, putKey, and deleteKey, respectively — which performs all of the logic for handling that request and interacting with the backing database. Granted, this is an incredibly simple application and doesn’t account for