知识卡片

普通读书笔记卡 · 2142.v

利用率与排队延迟冲突

理解等级:高级

资源同一时刻只能服务有限工作,使用率升高后任务更常等待,用户延迟随之上升。故严格响应时间下不能追求所有资源始终满载;高吞吐与低延迟之间需要选择工作点。该模型说明接近饱和时排队压力,曲线形状仍取决于实际到达和服务分布。

这张卡片看不懂?

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

    • 理解性能的心智模型 · 查看原书依据

      Only a single thread can use the most resources at a given time, which forces threads to wait and take turns. As the utilization of the system increases, the processes have to wait longer on average for resources to become available. This wait increases the latency that users experience. Because most applications must respond within latency bounds, they can’t afford to utilize the system entirely.

    参考来源

    - 位置:《Software Architecture and Decision-Making》第3章《理解性能的心智模型》(来源项目:book-sources;源文件:extracted/Software Architecture and Decision-Making Leveraging Leadership, Technology, and Product Management to Build Great Products… (Srinath Perera)/_epub-src/ch03.xhtml) - 结论依据:原文的定义、比较或例子支持“资源同一时刻只能服务有限工作,使用率升高后任务更常等待,用户延迟随之上升”;正文据此说明用途和边界。 - 原始内容:Only a single thread can use the most resources at a given time, which forces threads to wait and take turns. As the utilization of the system increases, the processes have to wait longer on average for resources to become available. This wait increases the latency that users experience. Because most applications must respond within latency bounds, they can’t afford to utilize the system entirely.