Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.
Case Abril: Click counterLuís Cipriani@lfcipriani (twitter, linkedin, github, ...)21o. GURU (2012-01-28) - Sao Paulo/Brazil
ME
motivation     • hard news     • sites Home must be dynamic     • knowing what’s hot helps a lot     • other advantages co...
architecture
architecture      • triggered with user click on the page
architecture
architecture          • servers click.js          • generate metrics dashboard
architecture
architecture          • clean up referer link list periodically          • avoid Redis out of memory
architecture
performance    • ~ 650.000 hits/day (only sites home)    •17 sites monitored    • ~ $823 dolars/month    • 15 heroku dynos...
performance• GET /click (on average day)• average: 6.8 ms• std deviation: 0.03 ms
overall performance
performance        • GET /dashboard (on average day)        • average: 4000 ms        • std deviation: 2.4 ms        • acc...
process performance  • continuous deploy  • few requirements = focus  • 3 developers working 1 month  • 2 weeks = initial ...
feedback (from emails) “A sua ferramenta é o que há de bom.” Edna Dantas (redatora chefe Contigo) “Parabéns a você e a sua...
Reformulação Box de Login                         Abril ID               http://engineering.abril.com.br/              ?
Upcoming SlideShare
Loading in …5
×

Case Abril: Tracking real time user behavior in websites Homes with Ruby, Sinatra, Heroku, Redis

7,251 views

Published on

This presentation shows how, in less than 1 month, the Brazilian publishing company Abril build a system to track user visiting in realtime to help publishers decide what would be the best content to atract more users to websites. This system uses ruby, sinatra, heroku and redis.

Published in: Technology, Business
  • Be the first to comment

Case Abril: Tracking real time user behavior in websites Homes with Ruby, Sinatra, Heroku, Redis

  1. 1. Case Abril: Click counterLuís Cipriani@lfcipriani (twitter, linkedin, github, ...)21o. GURU (2012-01-28) - Sao Paulo/Brazil
  2. 2. ME
  3. 3. motivation • hard news • sites Home must be dynamic • knowing what’s hot helps a lot • other advantages could be obtained
  4. 4. architecture
  5. 5. architecture • triggered with user click on the page
  6. 6. architecture
  7. 7. architecture • servers click.js • generate metrics dashboard
  8. 8. architecture
  9. 9. architecture • clean up referer link list periodically • avoid Redis out of memory
  10. 10. architecture
  11. 11. performance • ~ 650.000 hits/day (only sites home) •17 sites monitored • ~ $823 dolars/month • 15 heroku dynos handling everything
  12. 12. performance• GET /click (on average day)• average: 6.8 ms• std deviation: 0.03 ms
  13. 13. overall performance
  14. 14. performance • GET /dashboard (on average day) • average: 4000 ms • std deviation: 2.4 ms • acceptable by the publishers
  15. 15. process performance • continuous deploy • few requirements = focus • 3 developers working 1 month • 2 weeks = initial implementation using Mongo • 2 weeks = refactoring to Redis and improving dashboard load time
  16. 16. feedback (from emails) “A sua ferramenta é o que há de bom.” Edna Dantas (redatora chefe Contigo) “Parabéns a você e a sua equipe pelo desenvolvimento desta ferramenta. Ela é uma ajuda e tanto no nosso processo de publicação e edição do site.” Edna Dantas “Nós também estamos viciados no click counter. “ Hoshino (Webmaster Veja)
  17. 17. Reformulação Box de Login Abril ID http://engineering.abril.com.br/ ?

×