= 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 http://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 3.0+)
* 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, :forced_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 http://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 http://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: http://groups.google.com/group/db-charmer
- Subscribe using the info page or by sending an email to mailto: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 2.2, 2.3 and 3.0 using a few
different versions of Ruby. For more information about the build matrix please visit
http://github.com/kovyrin/db-charmer-sandbox web site.
In addition to CI testing, we use this gem 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, 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 3.0, 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
* Dmytro Shteflyuk
* Eric Lindvall
* Gregory Man
* Michael Birk
* Tyler McMullen