MongoDB Philly: MongoDB at Meeteor

Meeteor makes networking easier than ever by using your LinkedIn and Facebook data to help you discover the right people to meet. Founder and CTO Brandon Hilkert, will go through Meeteor's stack and showcase how MongoDB powers this professional recommendation engine.

Transcript

1. how uses Tuesday, April 10, 12
2. Brandon Hilkert @brandonhilkert http://meeteor.com Tuesday, April 10, 12
3. http://meeteor.com what is Tuesday, April 10, 12
4. EASY FASTand Tuesday, April 10, 12
5. we’re PRODUCT people, not SERVER people Tuesday, April 10, 12
6. Rails Nginx EC2 MySQL RDS platform Tuesday, April 10, 12
7. 2 pain points Tuesday, April 10, 12
8. 1 user signup Avg. 50 sec. Tuesday, April 10, 12
9. 1 user signup FAST? Avg. 50 sec. Tuesday, April 10, 12
10. 2 matching guru Tuesday, April 10, 12
11. Tuesday, April 10, 12
12. join spaghetti Tuesday, April 10, 12
13. DISCLAI MER! DISCLAIME R! DISCLAIME R! DISCLAIM E R! DISCLAIME R! Tuesday, April 10, 12
14. FASTER! Need to get Tuesday, April 10, 12
15. SQL vs. NoSQ L arly 2011 Tuesday, April 10, 12
16. flame wars are not always a BAD THING? Tuesday, April 10, 12
17. 1awareness Tuesday, April 10, 12
18. 2 o.s. s. Tuesday, April 10, 12
19. who CAREs? Tuesday, April 10, 12
20. Tuesday, April 10, 12
21. industrie s industrie s employer s employer s interests interests skills skills Tuesday, April 10, 12
22. document store great for comparisons if attributes are contained no more joins...sort of Tuesday, April 10, 12
23. mongo DB vs. couchD B Tuesday, April 10, 12
24. mongo DB winner! Tuesday, April 10, 12
25. data types Users id name email location 3 “Brandon” “b@m.com” “Philly” User_Interests id user_id interest_id 1 3 4 Interests id name 4 “HTML” m ysql many-to- many Tuesday, April 10, 12
26. data types Users id name email location interests 3 “Brandon” “b@m.com” “Philly” [“Ruby”, “MongoDB”] m ongo db more NATURAL? User.all(interests: “MongoDB”) array Tuesday, April 10, 12
27. embedded documents Users id name email location 3 “Brandon” “b@m.com” “Philly” User_Jobs id user_id job_id 1 3 4 Jobs id employer_id position description 4 7 “Developer” “Web” many-to- many Employers id name location industry 7 “Meeteor” “Philly” “Internet” m ysql Tuesday, April 10, 12
28. embedded documents m ongo db class User include MongoMapper::Document key :first_name, String key :last_name, String key :headline, String key :email, String many :jobs end class Job include MongoMapper::EmbeddedDocument key :employer, String key :position, String key :description, String key :industry, String end Tuesday, April 10, 12
29. embedded documents m ongo db User id name email location jobs user.jobs.first.employer => “Meeteor” 3 “Brandon” “b@m.com” “Philly” [{:_id => BSON::ObjectId('4e4efad0d052fc3ea0009a7a'), :employer => “Meeteor”, :position => “Founder”, :description => “Create networking application”, :industry => “Internet” }] embedded document Tuesday, April 10, 12
30. conversion? ~ 40 hours of which ~20 hours tinkering Tuesday, April 10, 12
31. RESUL TS user signup 50 sec. 3 sec. Tuesday, April 10, 12
32. RESUL TS matching guru 2 queries with a crap load of joins 2 SIMPLE queries Tuesday, April 10, 12
33. Rails Heroku MongoDB MongoHQ platform revise d we’re PRODUCT people, remember? Tuesday, April 10, 12
34. how to not make life miserable after converting... slow queries indexing db.system.profile.find({millis:{$gt:2000}}).sort({$natural:-1}) User.ensure_index(:interests) User.ensure_index("jobs.employer") AnalyticLog.ensure_index([[:user_id, 1], [:type, 1]]) Tuesday, April 10, 12
35. questions ? Brandon Hilkert @brandonhilkert Tuesday, April 10, 12