知识卡片
恢复动作可能形成正反馈
局部故障使请求变慢或容量下降,其他组件为补偿而重试或重分配流量,反过来使剩余容量更紧张,就形成级联失效。原书DynamoDB案例中恢复后的同时重连压垮元数据服务,说明“帮助恢复”的动作也可能扩大故障。应在边界限制这类放大,不能只看最初触发事件。
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
- Resilience · 查看原书依据
The mechanisms of cascading failures can vary a bit, but one thing they share is some kind of positive feedback mechanism. One part of a system experiences a local failure — a reduction in capacity, an increase in latency, etc. — that causes other components to attempt to compensate for the failed component in a
参考来源
- 位置:《Cloud Native Go》第09章《Resilience》(来源项目book-sources;源文件:extracted/Cloud Native Go (Matthew A. Titmus)/_epub-src/OEBPS/ch09.xhtml)
- 结论依据:局部故障使请求变慢或容量下降,其他组件为补偿而重试或重分配流量,反过来使剩余容量更紧张,就形成级联失效。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 mechanisms of cascading failures can vary a bit, but one thing they share is some kind of positive feedback mechanism. One part of a system experiences a local failure — a reduction in capacity, an increase in latency, etc. — that causes other components to attempt to compensate for the failed component in a