Analyze an nginx generated cache.log from http://openmymind.net/Caching-Your-Worst-Best-Friend/

cache_stats.rb

def to_route(url)
  #strip out the /v4/
  url = url[4..-1] 
  dot = url.index('.')
  #strip out the extension
  url = url[0...dot] unless dot.nil? 

  parts = url.split('/')
  if parts.length == 1
    "list #{parts[0]}"
  elsif parts.length == 2
    "show #{parts[0]}"
  elsif parts.length == 3
    "list #{parts[0]}/#{parts[2]}"
  elsif parts.length == 4
    "show #{parts[0]}/#{parts[2]}"
  else
    "?? #{url}"
  end
end

data = Hash.new {|h, k| h[k] = {hits: 0, misses: 0, total: 0, size: 0, time: 0}}

File.readlines('cache.log').each do |line|
  next unless line.start_with?('GET')
  parts = line.strip.split(' ')
  route = to_route(parts[1])
  d = data[route]
  cache = parts[2]

  d[:total] += 1
  d[:size] += parts[3].to_i
  if cache == 'hit'
    d[:hits] += 1
  elsif cache == 'miss'
    d[:misses] += 1
    d[:time] += (parts[4].to_f * 1000).to_i
  end
end

puts "route,total,hit ratio,size,miss time"
data.sort {|a, b| b[1][:total] <=> a[1][:total]}.each do |route, data|
  hit_ratio = data[:hits] * 100 / data[:total]
  # averages aren't great, but it works ok for us in this case
  # if your size or miss cost varies a lot, consider using something better
  avg_size = data[:size] / data[:total]
  avg_miss_time = data[:time] / data[:misses]

  puts "#{route},#{data[:total]},#{hit_ratio},#{avg_size},#{avg_miss_time}"
end