Scalability: how eBay handles it

Today, thanks to a talk on our user group, I've discovered this great pdf about eBay's architecture, presented by Randy Shoup and Dan Pritchett at SD Forum 2006.

The pdf gives a great overview on how eBay has evolved during the years, from a little application based on few Perl scripts to a big distributed application based on something like 15,000 application instances running in eight data centers.

The presented statistics are incredible:

  • 212,000,000 registered users
  • 1 billion page views per day
  • 26 billion SQL queries and updates per day
  • Over 2 petabytes of data
  • $1,590 worth of goods traded per second
  • Over 1 billion photos
  • 7 languages
  • 99.94% uptime

and about the development process:

  • Over 300 new features released each quarter
  • Over 100,000 lines of code released every two weeks
  • There's an interesting lesson that you can learn by reading the eBay case: the platform chosen (in this case J2EE) is not always sufficient to handle all the possible scalability problems (they have totally rewritten the J2EE connection pool for example), but you can start with an architecture and update it in the future if your business requirements will require so.

    It's an interesting reading...

    Print | posted on Friday, December 29, 2006 11:30 AM

    Comments on this post

    No comments posted yet.

    Your comment:

     (will show your gravatar)
     
    Please add 8 and 7 and type the answer here: