leaked google reader internal data

It’s known that Google doesn’t really like to let the public know much about their internal stuffs, like the number of servers they are using. However, one of their internal videos containing confidential information was “accidentally” uploaded to Google Video (now deleted).

Here’s are some “interesting facts” of Google Reader from Google Operating System:

  • Google Reader uses 10 TB for storing all the raw data
  • Google Reader crawls 8 million feeds
  • The rate of user growth = the rate of growth for the number of feeds
  • Search requires a lot of computational resources. Google Reader uses two indexes for search: a big tree updated twice a day (150 machines, 600 million documents) and 40 small trees for recent posts, updated every 5 minutes (40 machines, 40 million documents)
  • Some upcoming features are also mentioned in the video: internationalization, feed recommendations, and accepting pings sent to Google Blog Search
  • More on Google OS and Google Blogoscoped

I am rather surprised that they recorded it in video, and labeled it as Confidential (hands from AlexKing.org), like the word “curiosity” doesn’t exist.

[screenshot from Google Blogoscoped]