知识卡片
副本收益依赖独立故障
理想模型中,若N个副本的故障独立、任一存活即可提供服务且路由有效,整体可用性为1−各副本不可用概率的乘积。例如两个可用性0.99的独立副本得到1−0.01²=0.9999。共同机房、依赖故障或剩余容量不足会破坏假设,不能据副本个数承诺真实几个九。
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
- Resilience · 查看原书依据
This method generalizes to any number of components arranged in parallel, so that the availability of any N components is equal to 1 minus the product of their unavailabilities: A s = 1 - ∏ 1 N 1 - A i When all A i are equal, then this can be simplified to: A s
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:理想模型中,若N个副本的故障独立、任一存活即可提供服务且路由有效,整体可用性为1−各副本不可用概率的乘积。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:This method generalizes to any number of components arranged in parallel, so that the availability of any N components is equal to 1 minus the product of their unavailabilities: A s = 1 - ∏ 1 N 1 - A i When all A i are equal, then this can be simplified to: A s