benchmark.rb

require 'faker'
require 'factory_girl'
require 'test/factories'
require 'message'

Message.collection.drop
Message.create_indexes
Contact.collection.drop
Contact.create_indexes

def show_time_taken(message = "", &block)
  start_time = Time.zone.now
  yield
  duration_in_ms = 1000.0 * (Time.zone.now - start_time)
  time = Time.zone.now.strftime('%H:%M:%S %Y-%m-%d')
  puts ("%.1fms #{message}" % duration_in_ms)
end

class Array
  def rand_power_law(n = 1)
    min = 0
    max = size
    pl = ((max**(n+1) - min**(n+1))*Kernel.rand() + min**(n+1))**(1.0/(n+1))
    self[(max-1-pl.to_i)+min]
  end
end

batch_size = 1000
batches = 250

ten_thousands = (batch_size * batches) / 10_000

number_of_accounts = 50
contacts_per_account = 10

accounts = number_of_accounts.times.collect {Factory(:activated_account)}

account_msisdns = accounts.inject({}) do |memo, account|
  memo[account] = 100.times.collect { Faker::O2.msisdn }
  memo
end

puts "Starting benchmark to import #{batch_size * batches} messages"

show_time_taken "to process #{batches * batch_size}" do
  ten_thousands.times do |tens|
    puts "----------------------------------"
    show_time_taken "last 10k messages" do
      (batches / ten_thousands).times do |n|
        total = (batch_size * (n + 1)) + (tens * 10_000)
        show_time_taken "to process #{batch_size} messages (#{total} in total)" do
          batch_size.times do
            Rails.logger.info("--------")
            account = accounts.rand_power_law(2)
            contact_msisdn = account_msisdns[account].rand_power_law(2)
            account.import_message!(Factory.attributes_for(:message).merge(:contact_msisdn => contact_msisdn))
          end
        end
      end
    end
  end
end

db.rake

  task :benchmark => 'db:reset' do
    raise "This task can only run in development" unless Rails.env.development?
    benchmark_file = File.join(Rails.root, 'db', 'benchmark.rb')
    load(benchmark_file) if File.exist?(benchmark_file)
  end