Flickr Scalabilty

Loading...

Flash Player 9 (or above) is needed to view presentations.
We have detected that you do not have it on your computer. To install it, go here.

0 comments

Post a comment

    Post a comment
    Embed Video
    Edit your comment Cancel

    4 Favorites

    Flickr Scalabilty - Presentation Transcript

    1. Scalability - Flickr Shimon koifman
      • Place to share photos
      • One of the earliest web 2.0 applications
      • Online community platform
            • 470M photos, 4 or 5 sizes of each
            • More than 4 billion queries per day
            • 2 PB raw storage (consumed about ~1.5TB on Sunday)
            • Over 400,000 photos being added every day
          • Services
            • Atom/RSS/RDF Feeds
            • APIs
              • SOAP
              • XML-RPC
              • REST
              • PEAR::XML::Tree
        • Performance
        • High-Availability
        • Technology X
        • Protocol Y
        • Traffic growth
        • Dataset growth
        • Maintainability
        • Scale
        • High-Availability
        • Performance
        • The trade off’s between good/fast/cheap
        • Application servers scale in two ways:
          • Really well
          • Quite badly
        • Local session == bad
          • When they move == quite bad
        • Centralized sessions == good
        • No sessions == Great
        • Data size is at 12 TB of user metadata (these are not photos)
        • 16GB RAM
        • 6-disk 15K RPM
          • Snapshot of db1.flickr.com
            • SELECT’s 44,220,588
            • INSERT’s 1,349,234
            • UPDATE’s 1,755,503
            • DELETE’s 318,439
          • 13 SELECT’s per I/U/D
        • Database is the toughest part to scale
        • The best thing is to avoid the issue all together and buy a bigger hardware
        • Dual Opteron/Intel64 systems with 16+GB can get you a long way…
        • Web applications typically have a read/write ratio of 80/20
        • If we can scale read capacity, we can solve a lot of situations
        • Replication, Replication , Replication
        • A bunch of slave servers handle all the SELECT’s
        • A single master just handles I/U/D’s
        • Have to ensure consistency in the application logic
        • It can scale horizontally, at least for a while.
    2. Slave Farm DB1 Master I/U/D’s SELECT’s Search Slave Farm Search SELECT’s DB3 Main Search slave DB2 Main Slave
    3. [email_address] www.comverse.com

    + shimikoifshimikoif, 2 years ago

    custom

    827 views, 4 favs, 1 embeds more stats

    based on Cal Henderson presentation - compressed

    More info about this document

    © All Rights Reserved

    Go to text version

    • Total Views 827
      • 826 on SlideShare
      • 1 from embeds
    • Comments 0
    • Favorites 4
    • Downloads 30
    Most viewed embeds
    • 1 views on http://www.visualcv.com

    more

    All embeds
    • 1 views on http://www.visualcv.com

    less

    Flagged as inappropriate Flag as inappropriate
    Flag as inappropriate

    Select your reason for flagging this presentation as inappropriate. If needed, use the feedback form to let us know more details.

    Cancel
    File a copyright complaint
    Having problems? Go to our helpdesk?

    Categories