Posted on   |  

ORDINARY PEOPLE COULD CHOOSE TO BE EXTRAORDINARY.

Github Repos

Paper Overview

Tutorial

System Design

Posted on   |   In   |  

System Design

In a system design interview, you should at least do the following 4 tips:

  • Ask good questions: what feature to work on? how much to scale?
    • To figure out which feature he cares about and which feature he doesn’t, and make sure the finally feature set is small, then go deep into this feature set.
    • How much data you need to store in the database, how many requests per second(QPS) you need to handle, and what kind of latency is expected.
  • Don’t use buzzwords.
  • Clear and organized thinking.
  • Dirve discussions(80-20 rule): you should be talking 80% of the time and the interviewer should be talking 20% of the time.
Read more »

Posted on   |  

小白视角：一文读懂社长的TinyWebServer(Raw_Version)

TL;DR 感谢社长，社长牛逼。

目录

  • 1.什么是Web Server（网络服务器）
  • 2.用户如何与你的Web服务器进行通信
  • 3.Web服务器如何接收客户端发来的HTTP请求报文呢
  • 4.Web服务器如何处理以及响应接收到的HTTP请求报文呢
  • 5.数据库连接池是如何运行的
  • 6.什么是CGI校验
  • 7.生成HTTP响应并返回给用户
  • 8.服务器优化：定时器处理非活动链接
  • 9.服务器优化：日志
  • 10.压测（非常关键）
  • 11.这个服务器的不足在哪（希望大家可以献计献策）
  • 12.如何在此基础添加功能把社长的变成自己的（小声儿bb）
  • 参考资料

预备知识：通读一遍《Linux高性能服务器编程》— 游双著

本文将带你从本人的小白视角，从头到尾彻底理解社长的TinyWebServer项目，明白每个部分都发生了什么，你的服务器程序又是如何处理，如何响应（response）来自客户端的用户请求的（requests）。

Read more »

Posted on   |   In   |  

Importance Sampling

1.What is Importance Sampling

Importance Sampling(IS) is a general technique for estimating properties of a particular distribution $p$, while only having samples generated from a different distribution $q$ than the distribution of interest($p$).

And actually, IS is a variant of Monte Carlo approximation. In Monte Carlo approximation, we were extimating the expected value of a random variable $X$ by this sample mean sum:

But, there is a big assumption that we can efficiently draw samples from the true distribution $p$ of this random variable $X$. What if we can’t do that? Can we use some alternative distribution like $q$ and use the samples drawing from $q$ to correct the fact that we were drawing samples from the wrong distribution or get a better estimation than its true distribution $p$? There is where Importance sampling comes for.

Read more »