Flickr architecture

Cal Henderson of Ludicorp gave a presentation to the Vancouver PHP Association on September 9 about Flickr’s general architecture and the use of PHP. A zip of the PowerPoint presentation is available on Flickr’s site. I converted the PowerPoint presentation to a PDF file for easy viewing. (via MovableBlog Asides) Some interesting statistics on Flickr as of September 2004:
  • About 60,000 lines of PHP code.
  • About 60,000 lines of templates.
  • About 25,000 database transactions per second at peak.
  • 13 selects per insert/update/delete.
  • About 1,000 pages per second at peak.
  • They use Java for their node service and as a FTP daemon.
  • Posted
  • Updated at
  • Comments [4]


Commentary on "Flickr architecture":

  1. Casey Marshall on wrote:

    Thank you for posting a very insightful look into the design behind a site I really admire (wish I could have been at that presentation!). I am a little confused by some of the slides (perhaps b/c they are out of context).

    In what way is “InnoDB faster” (on the “FULLTEXT Indexes” slide)? Is it much faster for selects? Surely for I/U/D operations, InnoDB would be slower, since it has the overhead of transactions, right?

  2. dragonliquid on wrote:

    as far as I know
    till MySQL 5.1 InnoDB doesn’t support FULLTEXT indexing

  3. Janak Prajapati on wrote:

    Very useful contents..spcially the flickr architecture…but in pdf there is no explaination about the topic..can u post this pdf in details?.. thanks for posting such entry.

  4. Naren on wrote:

    very insightful. i was looking for more information clustering and scaling servers.