Analyze Redis data usage

analysis.rb

#!/usr/bin/env ruby

require 'csv'

class Level
  attr_reader :key, :parent, :bytesize, :total_bytesize, :children

  def initialize parent, key, bytesize = 0
    @parent = parent
    @key = key
    @bytesize = bytesize
    @total_bytesize = bytesize
    @children = {}
  end

  def eql? other
    other.is_a?(Level) && other.key == self.key
  end

  def hash
    self.key.hash * 3 + 13
  end

  def full_key
    self.parent.full_key + ":#{self.key}"
  end

  def to_str
    "<Item##{self.full_key} total_bytesize=#{self.total_bytesize}>"
  end
  alias to_s to_str

  def add child
    if self.children.key? child.key
      self.children[child.key].bytesize = child.bytesize
      return self.children[child.key]
    end
    self.children[child.key] = child.tap { |ch| add_bytesize ch.total_bytesize }
  end
  alias << add

  def print_tree output, indent: 0, depth: nil, threshold: nil
    output.puts "#{self.key || '<root>'}: #{self.total_bytesize}#{" (#{self.bytesize})" if self.bytesize > 0}"
    return if !depth.nil? && depth <= 0
    options = { indent: indent + 1, depth: depth.nil? ? nil : depth - 1, threshold: threshold }
    self.children.values.lazy
      .select { |child| threshold.nil? || child.total_bytesize >= threshold }
      .each do |child|
      output.print "#{' ' * indent}|- "
      child.print_tree(output, **options)
    end
  end

  def bytesize= value
    diff = value - self.bytesize
    @bytesize = value
    self.add_bytesize diff
  end

  protected

  def add_bytesize bytesize
    @total_bytesize += bytesize
    self.parent.add_bytesize bytesize if self.parent
  end
end

root = Level.new nil, nil

# ARGF should be a file generated with redis-rdb-tools with `-c memory` option.
# See: https://github.com/sripathikrishnan/redis-rdb-tools/tree/548b11ec3c81a603f5b321228d07a61a0b940159#generate-memory-report
CSV.new(ARGF, headers: true).each_with_index do |row|
  cursor = root

  tokens = row['key'].split(':')
  item_token = tokens.pop
  tokens.each { |token| cursor = cursor.add(Level.new(cursor, token)) }
  cursor << Level.new(cursor, item_token, row['size_in_bytes'].to_i)
end

root.print_tree STDOUT, threshold: 1024**2