YouTube技术架构

谈不上翻译，就是摘录

1 billion video views per day

1.Apache

2.Python

3.Linux (SuSe)

4.MySQL

5.psyco, a dynamic python->C compiler

6.lighttpd for video instead of Apache



       -->NetScaler-->Apache plus Lighttpd(for video/multiple process)-->Python-app-server(format html)->databases(MySQL to ...)

       |

Client |->CDN (popular content)-->Youtube Server(less popular)



Keep it simple and cheap

Keep a simple network path

Use commodity hardware

Use Simple common tools provided by os





MySQL -->database partioning with users assigned to different shards

spreads writes and reads

replica lag(备份延迟？) 0



colocation arrangement

Use 5 or 6 data centers plus the CDN

8.Images are replicated to different data centers using BigTable



Stall for time. Creative and risky tricks can help you cope in the short term while you work out longer term solutions

做最重要的事情

outsource service for short time to market

Keep it simple

Shard or partition

Constant interation on bottlenecks

Succeed as a team



FastCGI

1) CGI是WebServer与其他独立进程交互的工具，FastCGI常驻内存，不必每次都fork一个子进程，从而提升了效率





I found this information interesting:



    The application code is written mostly in Python (the web app is not the bottleneck… the database RPC is)

    They use Apache for page content and lighttpd for serving video

    Thumbnails are now served by Google’s BigTable

    They’re running SuSE Linux with MySQL

    HW RAID-10 across multiple disks was too slow. HW RAID-1 with SW RAID-0 was faster because the Linux I/O scheduler could see the multiple volumes and would therefore schedule more I/O

Citrix 05年收购了NetScaler (http server/load balancer/发布平台)

原文：http://highscalability.com/youtube-architecture