Architecture of LargeScale Websites

1
Architecture of Large-Scale Websites

• Yu Feng
• 06/08/2008

2
Agenda

• Static Content
• Image Servers
• Database
• Cache
• Mirror
• Load Balance

3
Static Content

• Use CMS (Content Management System)
• Can be managed easily (add/remove/modify)
• Channels

4
Static Content (contd)

• Make it static as much as possible
• Dynamic pages are always much slower
• Static pages are more cache-friendly
• The ways that high traffic sites do
• Netease, Sina, Mop
• Convert dynamic pages to static ones (BBS)

5
Image Servers

• Use separate image servers
• Images
• Small size
• Large amount
• Reduce the stress of the main server
• Can be optimized
• nginx
• lighttpd

6
Database

• Improve the design of your database
• Tables
• Tuning
• Your application

7
Database (contd)

• Use database clusters
• Oracle
• MySQL
• MySQL Replication Master/Slave)
• MySQL-Proxy

8
Cache

• Why caching is important?

9
Cache (contd)

• Apache
• Squid
• Ncache
• Memcached

10
Mirror

• Different entrance

11
Load Balance

• DNS
• What is a L4/L7 switch?

12
Load Balance (contd)

• Hardware
• F5, Aleton

13
Load Balance (contd)

• LVS
• HAProxy

14
The End

• Thanks