brillo

Brillo

Brillo is a Rails database scrubber and loader, useful for making lightweight copies of your production database for development machines, with sensitive information obfuscated. Most configuration is done through YAML: Specify the models that you want to back up, what associations you want with them, and what fields should be obfuscated (and how).

Once that is done, dropping your local DB and replacing it with the latest scrubbed copy is as easy as rake db:load.

Under the hood we use Polo to explore the classes and associations you specify in brillo.yml, obfuscated fields as configured.

Installation

Add this line to your application's Gemfile:

gem 'brillo'

Generate a starter brillo.yml file and config/initializers/brillo.rb with

$ rails g brillo_config

If you're using Capistrano, add Brillo's tasks to your Capfile:

# Capfile
require 'capistrano/brillo'

Lastly, since the scrubber is pretty resource intensive you may wish to ensure it runs on separate hardware from your app servers:

# config/deploy.rb
set :brillo_role, :my_batch_role

Usage

Here's an example brillo.yml for IMDB:

name: imdb # Namespace the scrubbed file will occupy in S3
compress: true # Compresses the file after scrubbing (default: true)
explore:
  user: # Name of ActiveRecord class in snake_case
    tactic: all # Scrubbing tactic to use (see Brillo:TACTICS for choices)
    associations: # Associations to include in the scrub (ALL associated records included)
      - comments
  movie:
    tactic: latest # The latest tactic explores the most recent 1,000 records
    associations:
      - actors
      - ratings
  admin/note: # Corresponds to the Admin::Note class
    tactic: all
obfuscations: #
  user.name: name # Scrub user.name with the "name" scrubber (see Brillo::SCRUBBERS for choices)
  user.phone: phone
  user.email: email

Brillo uses the official aws-sdk to communicate with S3. There are a number of ways to pass your S3 credentials, but the simplest is to set AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY in your environment.

If you'd like to see the gem in use, check out the /example_app directory.

Creating a scrubbed copy of the database in production

$ rake db:scrub

Loading a database in development

$ rake db:load

Loading a database on a stage

$ cap staging db:load

Advanced Configuration

If the built in record selection tactics aren't enough for you, or you need a custom obfuscation strategy, you can add them via the initializer. They are available in the YAML config like any other strategy.

# config/initializers/brillo.rb

Brillo.configure do |config|
  config.add_tactic :oldest, -> (klass) { klass.order(created_at: :desc).limit(1000) }

  config.add_obfuscation :remove_ls, -> (field) {
    field.gsub(/l/, "X")
  }

  # If you need the context of the entire record being obfuscated, it is available in the second argument
  config.add_obfuscation :phone_with_id, -> (field, instance) {
    (555_000_0000 + instance.id).to_s
  }

  # In addition to setting your S3 credentials via env you can set them something like this
  config.transfer_config.secret_access_key  = Rails.application.secrets.secret_access_key
  config.transfer_config.access_key_id      = Rails.application.secrets.access_key_id
end

To Do

• Support alternative transfer mechanisms