I like making things go! At the moment, I'm SVP of Infrastructure and Operations at Etsy, and I'm currently pursuing a Master's degree in Human Factors and Systems Safety at Lund University.

15 Comments

  1. James Wheare   •  

    Nice slides, do you know if this talk was filmed/recorded?

  2. allspaw   •     Author

    It was! I’m not 100% sure when (or where) the video will be up, but I’ll ask the O’Reilly folks and post an update here. Thanks a lot!

  3. Steve Conover   •  

    Excellent talk, my favorite at the conference so far, thank you. The duet format was a real success too.

  4. Steve Conover   •  

    Do you know if the conferece folks are going to post the video?

    Also I think Paul mentioned that you’ve made it easy for developers to produce their own stats via gmetric…could you briefly explain what you do here?

    Thanks -Steve

  5. allspaw   •     Author

    Steve:

    Yeah…so it’s about as for a number of backend applications, developers can write a basic ‘stats’ file with metrics/counts about the application, something like:

    images_processed=12
    external_http_fetches=15
    etc.

    we then have a process that can come by and slurp those values and metric names into ganglia. We also use the same format to feed nagios and check against a similarly-formatted file which define low and high-water marks to alert on.

    As long as those stats files exist in a common location and follow that format, devs can add or subtract to those metrics themselves and know that they’ll be automatically pulled into the various monitoring and metrics goodness. No need to write yet another crapload of extra metric-specific scripts.

  6. Steve Conover   •  

    On another topic – could you give a general outline of how your deploy is accomplished? How is the actual update of servers accomplished, how do deploys that cause structural changes to the data change things? Just curious.

    -Steve

  7. Steve Conover   •  

    “how do deploys that cause structural changes to the data change things”

    to be clearer, are deploys involving structural data changes treated specially in any way, or are these like any other “10 deploys per day” change?

  8. Steve Conover   •  

    “As long as those stats files exist in a common location and follow that format, devs can add or subtract to those metrics themselves and know that they’ll be automatically pulled into the various monitoring and metrics goodness. No need to write yet another crapload of extra metric-specific scripts.”

    Very interesting, thanks for sharing.

  9. Pingback: Velocity 2009 « Joel Weierman

  10. Pingback: DBA survival BLOG » Blog Archive » A great talk

  11. James Kavanagh   •  

    Hi John,

    Great share, thank you. Really insightful.
    Is there any chance I could get a copy of the original video? We’d really like to show this to some management folks but only the second part that concerns culture.

    James

  12. allspaw   •     Author

    Hey James – I don’t actually have a copy of the original video. O’Reilly had it sent to blip.tv and I think only they have the original, which is a bummer.

  13. Pingback: Dev and Ops Cooperation at Flickr | Just another software blog

  14. Pingback: DevOps: The Past, The Present and the Future - Part One MidVision

  15. Pingback: Developers working in Production. Of course! Maybe, sometimes. What, are you nuts?Lexique Informatique

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>