• Share
  • Email
  • Embed
  • Like
  • Save
  • Private Content
High Performance Weibo QCon Beijing 2011
 

High Performance Weibo QCon Beijing 2011

on

  • 59,255 views

Sina Weibo is the most popular Microblogging platform in China. It has more than 100 million user and tens of millions of daily updates. This slide explains the performance challenges in Weibo ...

Sina Weibo is the most popular Microblogging platform in China. It has more than 100 million user and tens of millions of daily updates. This slide explains the performance challenges in Weibo platform.

Statistics

Views

Total Views
59,255
Views on SlideShare
18,245
Embed Views
41,010

Actions

Likes
158
Downloads
1,865
Comments
3

64 Embeds 41,010

http://timyang.net 34647
http://www.mysqlops.com 3139
http://www.cnblogs.com 2088
http://blog.caiwangqin.com 231
http://www.zhuaxia.com 121
http://reader.youdao.com 94
http://xianguo.com 85
http://yunti.sinaapp.com 68
http://yananguo.gotoip4.com 66
http://static.slidesharecdn.com 63
http://blog.bypat.com 60
http://zhuaxia.com 57
http://www.chinatech.us 39
http://www.php163.com 26
http://flavors.me 26
http://cache.baidu.com 22
http://www.techgig.com 19
http://www.zhangxiaona.net 19
http://www.cqumzh.cn 15
http://www.dcshi.com 15
http://www.mysqlops.com. 11
http://www.w3ccollege.org 10
http://twitter.com 10
http://translate.googleusercontent.com 8
http://webcache.googleusercontent.com 7
https://broadvision.clearvale.com 6
http://www.mysqlops.net 4
https://twitter.com 3
http://paper.li 3
http://us-w1.rockmelt.com 3
http://chinaonrails.com 3
http://www.corponline.org 3
http://210.51.190.51:8002 3
http://old.xianguo.com 2
https://www.linkedin.com 2
http://www.16kan.com 2
http://feeds.feedburner.com 2
http://timyang.net. 2
http://www.techgig.timesjobs.com 1
http://cache.baiducontent.com 1
http://www.wumii.com 1
http://web-01.mysqlops.com 1
http://localhost 1
http://kb.cnblogs.com 1
http://archive.cnblogs.com 1
http://corponline.org 1
http://mysqlops.net 1
file:// 1
http://mailreader.163.com 1
http://127.0.0.1 1
More...

Accessibility

Categories

Upload Details

Uploaded via as Apple Keynote

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel

13 of 3 previous next Post a comment

  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Processing…
  • Nice post and thanks for shareing and know better how weibo operate
    Are you sure you want to
    Your message goes here
    Processing…
  • deep and simple
    Are you sure you want to
    Your message goes here
    Processing…
  • 新浪微博架构和经验
    Are you sure you want to
    Your message goes here
    Processing…
Post Comment
Edit your comment
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • 细化索引拆分\n
  • \n
  • \n
  • \n
  • \n
  • \n
  • 比较图表\n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • json及性能对比\n
  • \n
  • \n
  • \n
  • 相同空间,5倍容量\n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • 增加图表等接口成果\n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • 迅速定位\n
  • \n
  • \n
  • \n
  • \n

High Performance Weibo QCon Beijing 2011 High Performance Weibo QCon Beijing 2011 Presentation Transcript

  • Build High Performance Weibo System @TimYang
  • • • 140 • •
  • • QCon 2010• WDC 2010
  • Agenda3 42 1
  • Part 1
  • ( )
  • MySQL••
  • • hash•• user_timeline index
  • •• • • 5•
  • MySQL + cache••• cache • •
  • NoSQL“Databases are specializing – the “onesize fits all” approach no longer applies.” MongoDB
  • NoSQL?• • Redis • MongoDB• • Cassandra
  • • MongoDB• Redis• HBase• Cassandra
  • Redis -• • snapshot - • vm - • diskstore - • aof -
  • Redis -• string: key value redisObject 16 bytes/item• list: 40 bytes / item• hash: zipmap(<64)• set/sorted set
  • Redis - Replication• rdb•
  • Redis -
  • Redis -•• failover• list/set (optional)
  • • MySQL• NoSQL MySQL cache
  • Part 2
  • ( )
  • → →
  • Web cache cache•
  • “Web json cache json ( ) 2~5K , xml 10k , protobuf 500 ”
  • RDBMS→Key value (JSON)→Protocol buffers(binary)
  • JSON• DB• Cache• Message Queue• API
  • PB• Numeric: varint, from 1 byte•• Java, C++, Python...
  • •• cache • “192.168.0.1” → “0xc0a80000” → varint
  • Benchmark Text Text( http://code.google.com/p/thrift-protobuf-compare/wiki/Benchmarking)
  • “We would like to provide public APIs thataccept protocol buffers as well as XML, bothbecause it is more efficient and because werejust going to convert that XML to protocolbuffers on our end anyway.” - Google
  • •••
  • “ ”
  • • LAMP• 1ms•
  • • MQ stat• MQ Processor stat
  • ?“1 ”
  • • Timeline•• • cache
  • 1.2.3.4. cache master5. replication
  • • 5
  • • 5•
  • • 5• •
  • • 5• • •
  • • RAM is the new disk•
  • “Percona Server now both SQL and NOSQLHandlerSocket,100 rps 12cores/24threads and380GB of RAM, mysqlRedis NoSQL ”
  • @jackbillow “ 1. PK lookup 2. cache ”@kobe “ innodb ”@TimYang “InnoDB Adaptive Hash Indexeshash ”
  • • NoSQL • binlog Redis • jbinlog https://github.com/tangfl/jbinlog
  • Part 3
  • app
  • • • • spam••
  • Text
  • •• GZIP•
  • Part 4
  • “@ Zheng:“ ” //@ :@bian " " http://t.cn/h0k4r (49)| (15)” 2 14 10:36 (22) | (10)
  • ••
  • • • • •
  • • • • • cache•
  • ••
  • Timeline Memcached MySQL Redis
  • •• RAM•
  • Q&A@TimYang