DB Charmer - ActiveRecord Connection Magic Plugin¶ ↑

DbCharmer is a simple yet powerful plugin for ActiveRecord that significantly extends its ability to work with multiple databases and/or database servers. The major features we add to ActiveRecord are:

1. Simple management for AR model connections (switch_connection_to method)

2. Switching of default AR model connections to separate servers/databases

3. Ability to easily choose where your query should go (Model.on_* methods family)

4. Automated master/slave queries routing (selects go to a slave, updates handled by the master).

5. Multiple database migrations with very flexible query routing controls.

6. Simple database sharding functionality with multiple sharding methods (value, range, mapping table).

For more information on the project, you can check out our web site at kovyrin.github.com/db-charmer.

Installation¶ ↑

There are two options when approaching DbCharmer installation:

• using the gem (recommended and the only way of using it with Rails 3.2+)

• install as a Rails plugin (works in Rails 2.x only)

To install as a gem, add this to your Gemfile:

gem 'db-charmer', :require => 'db_charmer'

To install DbCharmer as a Rails plugin use the following command:

./script/plugin install git://github.com/kovyrin/db-charmer.git

Notice: If you use DbCharmer in a non-rails project, you may need to set DbCharmer.env to a correct value before using any of its connection management methods. Correct value here is a valid database.yml first-level section name.

Easy ActiveRecord Connection Management¶ ↑

As a part of this plugin we’ve added switch_connection_to method that accepts many different kinds of db connections specifications and uses them on a model. We support:

1. Strings and symbols as the names of connection configuration blocks in database.yml.

2. ActiveRecord models (we’d use connection currently set up on a model).

3. Database connections (Model.connection)

4. Nil values to reset model to default connection.

Sample code:

class Foo < ActiveRecord::Model; end

Foo.switch_connection_to(:blah)
Foo.switch_connection_to('foo')
Foo.switch_connection_to(Bar)
Foo.switch_connection_to(Baz.connection)
Foo.switch_connection_to(nil)

Sample database.yml configuration:

production:
  blah:
    adapter: mysql
    username: blah
    host: blah.local
    database: blah

  foo:
    adapter: mysql
    username: foo
    host: foo.local
    database: foo

The switch_connection_to method has an optional second parameter should_exist which is true by default. This parameter is used when the method is called with a string or a symbol connection name and there is no such connection configuration in the database.yml file. If this parameter is true, an exception would be raised, otherwise, the error would be ignored and no connection change would happen.

This is really useful when in development mode or in a tests you do not want to create many different databases on your local machine and just want to put all your tables in a single database.

Warning: All the connection switching calls would switch connection only for those classes the method called on. You can’t call the switch_connection_to method and switch connection for a base class in some hierarchy (for example, you can’t switch AR::Base connection and see all your models switched to the new connection, use the classic establish_connection instead).

Multiple DB Migrations¶ ↑

In every application that works with many databases, there is need in a convenient schema migrations mechanism.

All Rails users already have this mechanism - rails migrations. So in DbCharmer, we’ve made it possible to seamlessly use multiple databases in Rails migrations.

There are two methods available in migrations to operate on more than one database:

1. Global connection change method - used to switch whole migration to a non-default database.

2. Block-level connection change method - could be used to do only a part of a migration on a non-default db.

Migration class example (global connection rewrite):

class MultiDbTest < ActiveRecord::Migration
   db_magic :connection => :second_db

   def self.up
     create_table :test_table, :force => true do |t|
       t.string :test_string
       t.timestamps
     end
   end

   def self.down
     drop_table :test_table
   end
 end

Migration class example (block-level connection rewrite):

class MultiDbTest < ActiveRecord::Migration
  def self.up
    on_db :second_db do
      create_table :test_table, :force => true do |t|
        t.string :test_string
        t.timestamps
      end
    end
  end

  def self.down
    on_db :second_db { drop_table :test_table }
  end
end

Migration class example (global connection rewrite, multiple connections with the same table): (NOTE: both :connection and :connections can take an array of connections)

class MultiDbTest < ActiveRecord::Migration
  db_magic :connections => [:second_db, :default]

  def self.up
    create_table :test_table, :force => true do |t|
      t.string :test_string
      t.timestamps
    end
  end

  def self.down
    drop_table :test_table
  end
end

Default Migrations Connection¶ ↑

Starting with DbCharmer version 1.6.10 it is possible to call ActiveRecord::Migration.db_magic and specify default migration connection that would be used by all migrations without excplicitly switched connections. If you want to switch your migration to the default ActiveRecord connection, just use db_magic :connection => :default.

Invalid Connection Names Handling¶ ↑

By default in all environments on_db and db_magic statments would fail if specified connection does not exist in database.yml. It is possible to make DbCharmer ignore such situations in non-production environments so that rails would create the tables in your single database (especially useful in test databases).

This behaviour is controlled by the DbCharmer.connections_should_exist configuration attribute which could be set from a rails initializer.

Warning: if in test environment you use separate connections and master-slave support in DbCharmer, make sure you disable transactional fixtures support in Rails. Without this change you’re going to see all kinds of weird data visibility problems in your tests.

Using Models in Master-Slave Environments¶ ↑

Master-slave replication is the most popular scale-out technique in a medium-sized and large database-centric applications today. There are some rails plugins out there that help developers to use slave servers in their models but none of them were flexible enough for us to start using them in a huge application we work on.

So, we’ve been using ActsAsReadonlyable plugin for a long time and have made tons of changes in its code over the time. But since that plugin has been abandoned by its authors, we’ve decided to collect all of our master-slave code in one plugin and release it for rails 2.2+. DbCharmer adds the following features to Rails models:

Auto-Switching all Reads to the Slave(s)¶ ↑

When you create a model, you could use db_magic :slave => :blah or db_magic :slaves => [ :foo, :bar ] commands in your model to set up reads redirection mode when all your find/count/exist/etc methods will be reading data from your slave (or a bunch of slaves in a round-robin manner). Here is an example:

class Foo < ActiveRecord::Base
  db_magic :slave => :slave01
end

class Bar < ActiveRecord::Base
  db_magic :slaves => [ :slave01, :slave02 ]
end

Default Connection Switching¶ ↑

If you have more than one master-slave cluster (or simply more than one database) in your database environment, then you might want to change the default database connection of some of your models. You could do that by using db_magic :connection => :foo call from your models. Example:

class Foo < ActiveRecord::Base
  db_magic :connection => :foo
end

Sample model on a separate master-slave cluster (so, separate main connection + a slave connection):

class Bar < ActiveRecord::Base
  db_magic :connection => :bar, :slave => :bar_slave
end

Per-Query Connection Management¶ ↑

Sometimes you have select queries that you know you want to run on the master. This could happen for example when you have just added some data and need to read it back and not sure if it made it all the way to the slave yet or no. For this situation and a few others there is a set of methods we’ve added to ActiveRecord models:

1) on_master - this method could be used in two forms: block form and proxy form. In the block form you could force connection switch for a block of code:

User.on_master do
  user = User.find_by_login('foo')
  user.update_attributes!(:activated => true)
end

In the proxy form this method could be used to force one query to be performed on the master database server:

Comment.on_master.last(:limit => 5)
User.on_master.find_by_activation_code(code)
User.on_master.exists?(:login => login, :password => password)

2) on_slave - this method is used to force a query to be run on a slave even in situations when it’s been previously forced to use the master. If there is more than one slave, one would be selected randomly. Tis method has two forms as well: block and proxy.

3) on_db(connection) - this method is what makes two previous methods possible. It is used to switch a model’s connection to some db for a short block of code or even for one statement (two forms). It accepts the same range of values as the switch_connection_to method does. Example:

Comment.on_db(:olap).count
Post.on_db(:foo).find(:first)

By default in development and test environments you could use non-existing connections in your on_db calls and rails would send all your queries to a single default database. In production on_db won’t accept non-existing names.

This behaviour is controlled by the DbCharmer.connections_should_exist configuration attribute which could be set from a rails initializer.

Forced Slave Reads¶ ↑

In some cases we could have models that are too important to be used in default “send all reads to the slave” mode, but we still would like to be able to switch them to this mode sometimes. For example, you could have User model, which you would like to keep from lagging with your slaves because users do not like to see outdated information about their accounts. But in some cases (like logged-out profile page views, etc) it would be perfectly reasonable to switch all reads to the slave.

For this use-case starting with DbCharmer release 1.7.0 we have a feature called forced slave reads. It consists of a few separate small features that together make it really powerful:

1) :force_slave_reads => false option for ActiveRecord‘s db_magic method. This option could be used to disable automated slave reads on your models so that you could call on_slave or use other methods to enable slave reads when you need it. Example:

class User < ActiveRecord::Base
  db_magic :slave => slave01, :force_slave_reads => false
end

2) force_slave_reads ActionController class method. This method could be used to enable per-controller (when called with no arguments), or per-action (:only and :except params) forced reads from slaves. This is really useful for actions in which you know you could tolerate some slave lag so all your models with slaves defined will send their reads to slaves. Example:

class ProfilesController < Application
  force_slave_reads :except => [ :login, :logout ]
  ...
end

3) force_slave_reads! ActionController instance method, that could be used within your actions or in controller filters to temporarily switch your models to forced slave reads mode. This method could be useful for cases when the same actions could be called by logged-in and anonymous users. Then you could authorize users in before_filter and call force_slave_reads! method for anonymous page views.

class ProfilesController < Application
  before_filter do
    force_slave_reads! unless current_user
  end
  ...
end

Notice: Before using this method you need to enable ActionController support in DbCharmer. You need to call DbCharmer.enable_controller_magic! method from your project initialization code.

4) DbCharmer.force_slave_reads method that could be used with a block of ruby code and would enable forced slave reads mode until the end of the block execution. This is really powerful feature allowing high granularity in your control of forced slave reads mode. Example:

DbCharmer.force_slave_reads do
  ...
  total_users = User.count
  ...
end

Notice: At this point the feature considered beta and should be used with caution. It is fully covered with tests, but there still could be unexpected issues when used in real-world applications.

Associations Connection Management¶ ↑

ActiveRecord models can have an associations with each other and since every model has its own database connections, it becomes pretty hard to manage connections in a chained calls like User.posts.count. With a class-only connection switching methods this call would look like the following if we’d want to count posts on a separate database:

Post.on_db(:olap) { User.posts.count }

Apparently this is not the best way to write the code and we’ve implemented an on_* methods on associations as well so you could do things like this:

@user.posts.on_db(:olap).count
@user.posts.on_slave.find(:title => 'Hello, world!')

Notice: Since ActiveRecord associations implemented as proxies for resulting objects/collections, it is possible to use our connection switching methods even without chained methods:

@post.user.on_slave - would return post's author
@photo.owner.on_slave - would return photo's owner

Starting with DbCharmer release 1.4 it is possible to use prefix notation for has_many and HABTM associations connection switching:

@user.on_db(:foo).posts
@user.on_slave.posts

Named Scopes Support¶ ↑

To make it easier for DbCharmer users to use connections switching methods with named scopes, we’ve added on_* methods support on the scopes as well. All the following scope chains would do exactly the same way (the query would be executed on the :foo database connection):

Post.on_db(:foo).published.with_comments.spam_marked.count
Post.published.on_db(:foo).with_comments.spam_marked.count
Post.published.with_comments.on_db(:foo).spam_marked.count
Post.published.with_comments.spam_marked.on_db(:foo).count

And now, add this feature to our associations support and here is what we could do:

@user.on_db(:archive).posts.published.all
@user.posts.on_db(:olap).published.count
@user.posts.published.on_db(:foo).first

Bulk Connection Management¶ ↑

Sometimes you want to run code where a large number of tables may be used, and you’d like them all to use an alternate database. You can now do this:

DbCharmer.with_remapped_databases(:logs => :big_logs_slave) { ... }

Any model whose default database is :logs (e.g., db_charmer :connection => :logs) will now have its connection switched to :big_logs_slave in that block. This is lower precedence than any other DbCharmer method, so Model.on_db(:foo).find(...) and such things will still use the database they specify, not the one that model was remapped to.

You can specify any number of remappings at once, and you can also use :master as a database name that matches any model that has not had its connection set by DbCharmer at all.

Note: DbCharmer works via alias_method_chain in model classes. It is very careful to only patch the models it needs to. However, if you use with_remapped_databases and remap the default database (:master), then it has no choice but to patch all subclasses of ActiveRecord::Base. This should not cause any serious problems or any big performance impact, but it is worth noting.

Simple Sharding Support¶ ↑

Starting with the release 1.6.0 of DbCharmer we have added support for simple database sharding to our ActiveRecord extensions. Even though this feature is tested in production, we do not recommend using it in your applications without complete understanding of the principles of its work.

At this point we support four sharding methods:

1) range - really simple sharding method that allows you to take a table, slice is to a set of smaller tables with pre-defined ranges of primary keys and then put those smaller tables to different databases/servers. This could be useful for situations where you have a huge table that is slowly growing and you just want to keep it simple and split the table load into a few servers without building any complex sharding schemes.

2) hash_map - pretty simple sharding method that allows you to take a table and slice it to a set of smaller tables by some key that has a pre-defined key of values. For example, list of US mailing addresses could be sharded by states, where you’d be able to define which states are stored in which databases/servers.

3) db_block_map - this is a really complex sharding method that allows you to shard your table into a set of small fixed-size blocks that then would be assigned to a set of shards (databases/servers). Whenever you would need an additional blocks they would be allocated automatically and then balanced across the shards you have defined in your database. This method could be used to scale out huge tables with hundreds of millions to billions of rows and allows relatively easy re-sharding techniques to be implemented on top.

4) db_block_group_map - really similar to the db_block_map method with a single difference: this method allows you to have a set of databases (table groups) on each server and every group would be handled as a separate shard of data. This approach is really useful for pre-sharding of your data before scaling your application out. You can easily start with one server, having 10-20-50 separate databases, and then move those databases to different servers as you see your database outgrow one machine.

How to enable sharding?¶ ↑

To enable sharding extensions you need to take a few things:

1) Create a Rails initializer (on run this code when you initialize your script/application) with a set of sharded connections defined. Each connection would have a name, sharding method and an optional set of parameters to initialize the sharding method of your choice.

2) Specify sharding connection you want to use in your models.

3) Specify the shard you want to use before doing any operations on your models.

For more details please check out the following documentation sections.

Sharded Connections¶ ↑

Sharded connection is a simple abstractions that allows you to specify all sharding parameters for a cluster in one place and then use this centralized configuration in your models. Here are a few examples of sharded connections initizlization calls:

1) Sample range-based sharded connection:

TEXTS_SHARDING_RANGES = {
  0...100   => :shard1,
  100..200  => :shard2,
  :default  => :shard3
}

DbCharmer::Sharding.register_connection(
  :name => :texts,
  :method => :range,
  :ranges => TEXTS_SHARDING_RANGES
)

2) Sample hash map sharded connection:

SHARDING_MAP = {
  'US'  => :us_users,
  'CA'  => :ca_users,
  :default  => :other_users
}

DbCharmer::Sharding.register_connection(
  :name => :users,
  :method => :hash_map,
  :map => SHARDING_MAP
)

3) Sample database block map sharded connection:

DbCharmer::Sharding.register_connection(
  :name => :social,
  :method => :db_block_map,
  :block_size => 10000,                   # Number of keys per block
  :map_table => :event_shards_map,        # Table with blocks to shards mapping
  :shards_table => :event_shards_info,    # Shards connection information table
  :connection => :social_shard_info       # What connection to use to read the map
)

After your sharded connection is defined, you could use it in your models:

class Text < ActiveRecord::Base
  db_magic :sharded => {
    :key => :id,
    :sharded_connection => :texts
  }
end

class Event < ActiveRecord::Base
  set_table_name :timeline_events

  db_magic :sharded => {
    :key => :to_uid,
    :sharded_connection => :social
  }
end

Switching connections in sharded models¶ ↑

Every time you need to perform an operation on a sharded model, you need to specify on which shard you want to do it. We have a method for this which would look familiar for the people that use DbCharmer for non-sharded environments since it looks and works just like those per-query connection management methods:

Event.shard_for(10).find(:conditions => { :to_uid => 123 }, :limit => 5)
Text.shard_for(123).find_by_id(123)

There is another method that could be used with range and hash_map sharding methods, this method allows you to switch to the default shard:

Text.on_default_shard.create(:body => 'hello', :user_id => 123)

And finally, there is a method that allows you to run your code on each shard in the system (at this point the method is supported in db_block_map and db_block_group_map methods only):

Event.on_each_shard { |event| event.delete_all }

Defining your own sharding methods¶ ↑

It is possible with DbCharmer for the users to define their own sharding methods. You need to do a few things to implement your very own sharding scheme:

1) Create a class with a name DbCharmer::Sharding::Method::YourOwnName

2) Implement at least a constructor initialize(config) and a lookup instance method shard_for_key(key) that would return either a connection name from database.yml file or just a hash of connection parameters for rails connection adapters.

3) Register your sharded connection using the following call:

DbCharmer::Sharding.register_connection(
  :name => :some_name,
  :method => :your_own_name,    # your sharding method class name in lower case
  ... some additional parameters if needed ...
)

4) Use your sharded connection as any standard one.

Adding support for default shards in your custom sharding methods¶ ↑

If you want to be able to use on_default_shard method on your custom-sharded models, you need to do two things:

1) implement support_default_shard? instance method on your sharded class that would return true if you do support default shard specification and false otherwise.

2) implement :default symbol support as a key in your shard_for_key method.

Adding support for shards enumeration in your custom sharding methods¶ ↑

To add shards enumeration support to your custom-sharded models you need to implement an instance method shard_connections on your class. This method should return an array of sharding connection names or connection configurations to be used to establish connections in a loop.

Documentation/Questions¶ ↑

For more information about the library, please visit our site at kovyrin.github.com/db-charmer. If you need more defails on DbCharmer internals, please check out the source code. All the plugin’s code is ~100% covered with tests (located in a separate staging rails project at github.com/kovyrin/db-charmer-sandbox). The project has unit tests for all or at least the most actively used code paths.

If you have any questions regarding this project, you could contact the author using the DbCharmer Users Group mailing list:

• Group Info: groups.google.com/group/db-charmer

• Subscribe using the info page or by sending an email to db-charmer-subscribe@googlegroups.com

What Ruby and Rails implementations does it work for?¶ ↑

We have a continuous integration setup for this gem on with Rails 2.3, 3.0, 3.1 and 3.2 using a few different versions of Ruby. For more information about the build matrix please visit github.com/kovyrin/db-charmer-sandbox web site.

In addition to CI testing, this gem is used in production on Scribd.com (one of the largest RoR sites in the world) with Ruby Enterprise Edition and Rails 2.2, Rails 2.3, Sinatra and plain Rack applications.

Starting with version 1.7.0 we support Rails 3.0-3.2, but until further notice we consider it a beta-quality feature since we do not run any production software on this version of Rails. If you run your application on Rails 3.x and having any issues with the gem, please contact the author.

Who are the authors?¶ ↑

This plugin has been created in Scribd.com for our internal use and then the sources were opened for other people to use. Most of the code in this package has been developed by Oleksiy Kovyrin for Scribd.com and is released under the MIT license. For more details, see the LICENSE file.

Other contributors who have helped with the development of this library are (alphabetically ordered):

• Allen Madsen

• Andrew Geweke

• Ashley Martens

• Cauê Guerra

• David Dai

• Dmytro Shteflyuk

• Eric Lindvall

• Eugene Pimenov

• Jonathan Viney

• Gregory Man

• Michael Birk

• Tyler McMullen