From "Overnight" to "Real-time": A Two-Year NoSQL Case Study

1,412 views

Published on

Meteor Solutions integrates site and advertising analytics to provide major publishers and advertisers the ability to identify and reach their influential visitors with advertising, exclusive content and rewards. Eighteen months ago, Meteor was backed by a relational DB and struggling to keep up with volumes in a batch processing environment that was ill-suited to our graph oriented data model. Today, the service is backed by Cloudant, a distributed document store based on CouchDB, and provides deeper analytics in real-time. This transition enabled 10x growth and allowed us to open our technology to a much broader range of applications -- though not without some bumps along the way. This talk will cover:

Overview of our services and specific technical challenges
Overview of Cloudant/CouchDB, how we leverage it, and its relation to other SQL, NoSQL, and web technologies in our stack
Benefits we've seen and tradeoffs we have had to make
Operational lessons learned
Future plans: how NoSQL's possibilities and limitations influence business, product and operational decisions

Published in: Technology, Business
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
1,412
On SlideShare
0
From Embeds
0
Number of Embeds
1
Actions
Shares
0
Downloads
4
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

From "Overnight" to "Real-time": A Two-Year NoSQL Case Study

  1. 1. 2 years of nosql benjamin anderson @banjiewen meteor solutions, inc.
  2. 2. about meteor
  3. 3. our data
  4. 4. 2009pretty simple
  5. 5. why this was bad
  6. 6. enter couchdb/cloudant (and some redis)
  7. 7. nosql in production
  8. 8. what we’ve learned
  9. 9. solved initial problems
  10. 10. introduced new ones
  11. 11. problemsno documentation
  12. 12. problemsno documentation
  13. 13. problemsyou probably don’t need data model flexibility
  14. 14. problemsad-hoc data analysis is challenging
  15. 15. problems predictability
  16. 16. problems library support
  17. 17. problems not unsolvable!
  18. 18. overall, good stuff
  19. 19. good stuff rapid iteration
  20. 20. good stufffanatical customer support
  21. 21. good stuffgood people want to work on it
  22. 22. good stuffhorizontal scalability is no joke
  23. 23. good stuffsolves real problems
  24. 24. we’ve learned a lot
  25. 25. what we’ve learned pay for the support
  26. 26. what we’ve learned expect tradeoffs
  27. 27. what we’ve learned you probably don’t need it
  28. 28. thanks benjamin anderson @banjiewenmeteor solutions, inc.

×