是否有任何关于10000客户端/秒问题解决方案的现代评论 [英] Is there any modern review of solutions to the 10000 client/sec problem

问题描述

(通常称为C10K问题)

(Commonly called the C10K problem)

对于 c10k 问题的解决方案是否有更现代的评论(最新更新:9月2日2006)，专门针对Linux(epoll，signalfd，eventfd，timerfd ..)和libev或libevent之类的库?

Is there a more contemporary review of solutions to the c10k problem (Last updated: 2 Sept 2006), specifically focused on Linux (epoll, signalfd, eventfd, timerfd..) and libraries like libev or libevent?

讨论现代Linux服务器上所有已解决但仍未解决的问题的事情?

Something that discusses all the solved and still unsolved issues on a modern Linux server?

推荐答案

C10K问题通常假定您正在尝试优化单个服务器，但是正如您所引用的文章所指出的那样，硬件不再是瓶颈".因此，要采取的第一步是确保仅添加更多硬件是最简单，最便宜的.

The C10K problem generally assumes you're trying to optimize a single server, but as your referenced article points out "hardware is no longer the bottleneck". Therefore, the first step to take is to make sure it isn't easiest and cheapest to just throw more hardware in the mix.

如果我们有一个500美元的盒子每秒为X个客户提供服务，那么再购买500美元的盒子使我们的吞吐量提高一倍，而不是让一个知道要花费多少小时和多少钱的员工吞噬，这会更有效率.如何从原包装中挤出更多.当然，这是假定我们的应用程序对多服务器友好的，我们知道如何进行负载平衡等，等等.

If we've got a $500 box serving X clients per second, it's a lot more efficient to just buy another $500 box to double our throughput instead of letting an employee gobble up who knows how many hours and dollars trying to figure out how squeeze more out of the original box. Of course, that's assuming our app is multi-server friendly, that we know how to load balance, etc, etc...

这篇关于是否有任何关于10000客户端/秒问题解决方案的现代评论的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！