Bạn có biết Twitter lưu trữ dữ liệu trên database nào?

Mình có search được 1 chút thì có một bài trên quora có cùng ý tưởng như vầy. Ý tưởng là sử dụng rất nhiều loại DB khác nhau với công dụng khác nhau.

Their GitHub page is quite informative on this topic, if you don't mind digging around a bit:

• MySQL - Twitter uses MySQL heavily for primary storage of Tweets and Users, and maintains a custom fork that they recently open-sourced: 
  https://github.com/twitter/mysql
  . More information on the engineering blog: 
  http://engineering.twitter.com/2...
• FlockDB - This is Twitter's in-house graph database, which they use to store social graph information (following, etc). Ultimately it's built on MySQL, but it's still basically a proper database in its own right: 
  https://github.com/twitter/flockdb
• Memcached: Twitter uses a "heavily modified" fork of Memcached 1.4.4 which they call Twemcache. It's been in production for well over a year already (as of 7/2012): 
  https://github.com/twitter/twemc...
  . Also check out lots more info on the Engineering Blog: 
  http://engineering.twitter.com/2...
• Cassandra: Twitter has spent a lot of time experimenting with Cassandra. Ultimately this led to services like Snowflake for quickly generating unique identifiers: 
  https://github.com/twitter/snowf...
  . There's also a Ruby gem for Cassandra here: 
  https://github.com/twitter/cassa...
• Gizzard: This is an in-house scala framework for building custom distributed databases (with arbitrary storage technology), that underlies a number of the other systems discussed here. GitHub page here: 
  https://github.com/twitter/gizzard
• Apache Lucene: This isn't on the GitHub page, but has been talked about publicly on the engineering blog. The search index is now powered by Lucene, through a system they call Earlybird. See 
  http://engineering.twitter.com/2...
   and 
  http://engineering.twitter.com/2...
   for more detail.
• HBase and Hadoop: Twitter uses Hadoop and HBase heavily, although this also isn't clear from the GitHub page. Check out 
  Kevin Weil
  's "Elephant Bird" project on GitHub: 
  https://github.com/kevinweil/ele...
  , with more information on the blog: 
  http://engineering.twitter.com/2...
• Redis: Finally, there's some experimental timeline storage technology that was developed on Redis. It's unclear whether Redis is being used in production or not at this time, but see the Haplocheirus project: 
  https://github.com/twitter/haplo...

There's plenty more, and surely lots of technology that has not been publicly disclosed, but that's the majority of public information that's available on the topic.