知识卡片
优化常把瓶颈转移到别处
CPU、内存、磁盘I/O与网络I/O是常见受限资源。缓存可能用内存换计算或磁盘访问,外部缓存又增加网络通信;增加实例也可能使共享依赖成为瓶颈。判断优化不能只看目标指标下降,还要观察代价转移及并发协调成本。
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
those tend to come at the expense of one or more other resources… so you’ll eventually find your system to be bottlenecked again by another resource. For example, a database might avoid disk I/O bottlenecking by storing data in memory, or a service might scale up by adding more instances whose communication overhead puts additional
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:CPU、内存、磁盘I/O与网络I/O是常见受限资源。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:those tend to come at the expense of one or more other resources… so you’ll eventually find your system to be bottlenecked again by another resource. For example, a database might avoid disk I/O bottlenecking by storing data in memory, or a service might scale up by adding more instances whose communication overhead puts additional