20% drop in traffic. [Amazon] found that even very small delays would result in substantial and costly drops in revenue. Users really respond to speed 2006
service Very good Google Analytics Free of charge Less detail Your logs No data to third parties Harder to use Profiling Complex setup required Overhead
* from comments; # select * from users where id = 1; # select * from users where id = 2; ... >>> comments = Comment.objects.all(); >>> for c in comments.select_related(“user”): print c.user.name # select comments.*, users.* # from comments, users # where comments.user_id = users.id;
* from users; # select * from comments where user_id = 1; ... >>> users = User.objects.filter(id__lt=10) >>> for u in users.prefetch_related(“comments”): print len(u.comments.all()) # select * from users where id < 10; # select * from comments # where user_id in (1,2,3,4,5,6,7,8,9); ## Joins them in python
end-user response time is spent in the front end Combine Compress Cache Less is more Load slow things later Focus on making the important things faster