Skip to content

Command line tool for generating anonymized database for MySQL/MariaDB

License

Notifications You must be signed in to change notification settings

kibitan/masking

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

MasKING🤴

CircleCI Acceptance Test MySQL Status Acceptance Test MariaDB Status

codecov Maintainability CodeScene Code Health Gem Version

The command line tool for anonymizing database records by parsing a SQL dump file and build a new SQL dump file with masking sensitive/credential data.

Design Concept

KISS ~ keep it simple, stupid ~

No connection to the database, No handling files, Only dealing with stdin/stdout. ~ Do One Thing and Do It Well ~ inspired by Unix Philosophy.

No External Dependency

Depend on only pure language standard libraries, no external libraries

Quality of Code

Heavily inspired by TDD. please see detail in presentation below.

Presentation / Demo

presentation

demo / slide: Generate anonymised database with MasKING

Installation

gem install masking

Requirement

  • Ruby 2.6/2.7/3.0/3.1/3.2/3.3

Supporting RDBMS

  • MySQL: 5.7, 8.0, 8.1
  • MariaDB: 10.21, 10.31, 10.4, 10.5, 10.6, 10.71, 10.81, 10.91, 10.10, 10.11, 11.0, 11.1

Usage

  1. Setup configuration for anonymizing target tables/columns to masking.yml

    NOTE: the columns which doesn't mention here will be NOT anonymized, it stays as it is.

    # table_name:
    #   column_name: masked_value
    
    users:
      string: anonymized string
      email: anonymized+%{n}@example.com # %{n} will be replaced with sequential number
      integer: 12345
      float: 123.45
      boolean: true
      null_column: null
      date: 2018-08-24
      time: 2018-08-24 15:54:06
      binary_or_blob: !binary | # Binary Data Language-Independent Type for YAML™ Version 1.1: http://yaml.org/type/binary.html
        R0lGODlhDAAMAIQAAP//9/X17unp5WZmZgAAAOfn515eXvPz7Y6OjuDg4J+fn5
        OTk6enp56enmlpaWNjY6Ojo4SEhP/++f/++f/++f/++f/++f/++f/++f/++f/+
        +f/++f/++f/++f/++f/++SH+Dk1hZGUgd2l0aCBHSU1QACwAAAAADAAMAAAFLC
        AgjoEwnuNAFOhpEMTRiggcz4BNJHrv/zCFcLiwMWYNG84BwwEeECcgggoBADs=
      # When a column name is suffixed with `?`, the original NULL value will not be anonymized.
      # This option can be beneficial for simulating SQL execution that closely resembles the original data.
      nullable_string?: anonymized nullable %{n} string

    A value will be implicitly converted to a compatible type. If you prefer to explicitly convert, you could use a tag as defined in YAML Version 1.1

    not-date: !!str 2002-04-28

    String should be matched with MySQL String Type. Integer/Float should be matched with MySQL Numeric Type. Date/Time should be matched with MySQL Date and Time Type.

    NOTE: MasKING doesn't check actual schema's type from the dump. If you put incompatible value it will cause an error during restoring to the database.

  2. Dump database with anonymizing

    MasKING works with mysqldump --complete-insert

      mysqldump --complete-insert -u USERNAME DATABASE_NAME | masking > anonymized_dump.sql
  3. Restore from the anonymized dump file

      mysql -u USERNAME ANONYMIZED_DATABASE_NAME < anonymized_dump.sql

    Tip: If you don't need to have an anonymized dump file, you can directly insert it from the stream. It can be faster because it has less IO interaction.

      mysqldump --complete-insert -u USERNAME DATABASE_NAME | masking | mysql -u USERNAME ANONYMIZED_DATABASE_NAME

options

$ masking -h
Usage: masking [options]
    -c, --config=FILE_PATH           specify config file. default: masking.yml
    -v, --version                    version

Use case of anonymized (production) database

  • Analyzing production databases for BI, Machine Learning, troubleshooting with respecting GDPR

  • Stress test / Integration test

  • Performance optimization for slow query

    The analyzing slow query often needs a similar amount of records/cardinality with production, the anonymized database help to analyze and tune the slow query.

  • Simulating database migration

    Some schema migration locks table and it causes trouble during the execution. With a smaller amount of database, the migration will finish in a short time and easy to overlook the problem. With the anonymized production database, it is easy to simulate the migration as the real release and makes it easy to find the problem.

  • Better feature development flow

    Using similar data with the production database makes better development experience. It makes easy to find out the things which should be changed/fixed. Also, some bugs are related to unexpected data in production, it makes easy to find them too.

  • And… your idea here!

Development

git clone git@github.com:kibitan/masking.git
bin/setup

You can also run bin/console for an interactive prompt that will allow you to experiment.

To install this gem onto your local machine, run bundle exec rake install.

boot

  bundle exec exe/masking

Run test & rubocop & notes

  bundle exec rake

acceptance test

./acceptance/run_test.sh

available option via environment variable:

  • MYSQL_HOST: database host(default: localhost)
  • MYSQL_USER: mysql user name(default: mysqluser}
  • MYSQL_PASSWORD: password for user(default: password)
  • MYSQL_DBNAME: database name(default: mydb)

NOTE: run with TRACE=1 will show debug print. for the CI, TRACE environment variable on setting field in the repository

with docker-compose
docker-compose -f docker-compose.yml -f docker-compose/mysql80.yml run -e MYSQL_HOST=mysql80 app acceptance/run_test.sh

or

docker-compose/acceptance_test.sh mysql80

The docker-compose file names for other database versions, specify that file.

bundle exec mdl *.md

Development with Docker

docker build . -t masking
echo "sample stdout" | docker run -i masking
docker run masking -v
docker run --entrypoint sh -it masking # inside of docker container

Profiling

use bin/masking_profile

 $ cat your_sample.sql | bin/masking_profile
flat result is saved at /your/repo/profile/flat.txt
graph result is saved at /your/repo/profile/graph.txt
graph html is saved at /your/repo/profile/graph.html

 $ open profile/flat.txt

see also: ruby-prof/ruby-prof: ruby-prof: a code profiler for MRI rubies

Benchmark

use benchmark/run.rb

$ benchmark/run.rb
       user     system      total        real
   1.103012   0.009460   1.112472 (  1.123093)

Future Todo

  • Pluggable/customizable for a mask way e.g. integrate with Faker
  • Parse the schema information and validate the target columns value
  • Performance optimization
    • Write in the streaming process
    • rewrite by another language?
    • establish benchmark

Contributing

Bug reports and pull requests are welcome on GitHub at https://github.com/kibitan/masking. This project is intended to be a safe, welcoming space for collaboration, and contributors are expected to adhere to the Contributor Covenant code of conduct.

License

The gem is available as open source under the terms of the MIT License.

Code of Conduct

Everyone interacting in the Masking project’s codebases, issue trackers, chat rooms and mailing lists is expected to follow the code of conduct.

1: MariaDB 10.2, 10.3, 10.7, 10.8, 10.9 is already not supported by official