Which one is best for performance from 'order' and 'sort_by'?
activerecord, postgresql, ruby, ruby-on-rails, ruby-on-rails-4
Solution
Database processing is few orders of significance faster, than Ruby. And database processing scales exclusively well, whereas Ruby's processing slowdown is proportional to the increase of the size of data you are processing.
Processing with Ruby drastically increases both time and (especially) memory consumption, and it can easily overload the memory and never actually finish the processing having the dataset is "big".
Some calculations with 1_000_000 rows with Ruby would take few tens of seconds, whereas PostgreSQL would finish it within few seconds.
Problem
Which one is best for performance ? 1) order (Database side calculation) 2) sort_by (Ruby side calculation) As per my thought `sort_by` should be fast then `order`. Because `order` is perform from database side so it is slower and `sort_by` perform on ruby side after getting result but i am confusing by seeing this result ex. => Normal query (take more time then `order` and `sort_by` query) :- ``` > User.all SELECT "users".* FROM "users" #<Benchmark::Tms:0x000000089e4398 @cstime=0.0, @cutime=0.0, @label="", @real=6.515499990200624e-05, @stime=0.0, @total=0.0, @utime=0.0> ``` => Query using `order` with benchmark (take less time then normal query and `sort_by` query) :- ``` > User.order(:name) # `order` query is perform in sql so should take more time User Load (0.6ms) SELECT "users".* FROM "users" ORDER BY "users"."name" ASC #<Benchmark::Tms:0x00000007ec48c8 @cstime=0.0, @cutime=0.0, @label="", @real=0.00014305600052466616, @stime=0.0, @total=0.0, @utime=0.0> ``` => Query using `sort_by` with benchmark (take more then `order` query and less time then normal query) :- ``` > User.all.sort_by(&:name) # Not perform `order` from database side so should be faster then `order` SELECT "users".* FROM "users" #<Benchmark::Tms:0x0000000897e5c0 @cstime=0.0, @cutime=0.0, @label="", @real=0.0019228710007155314, @stime=0.0, @total=0.0, @utime=0.0> ``` Conclusion :- Normal query real time :- 6.515499990200624e-05 `order` query real time :- 0.00014305600052466616 `sort_by` query real time :- 0.0019228710007155314 From this conclusion can we say that :- It's take too much time for normal query then applying `order` query. we have to use `order` method to decrease query perform time ? I always prefer `sort_by` as per my thinking. Which one is the best :(