= ActiveRecord::Turntable {}[http://travis-ci.org/drecom/activerecord-turntable] {}[https://gemnasium.com/drecom/activerecord-turntable] {}[https://coveralls.io/r/drecom/activerecord-turntable]
ActiveRecord::Turntable is a database sharding extension for ActiveRecord.
== Dependencies
activerecord(>=3.0.0)
== Supported Database
Currently supports mysql only.
== Installation
Add 'activerecord-turntable' to Gemfile:
gem 'activerecord-turntable', '~>1.1.0'
Run a bundle install:
bundle install
Run install generator:
bundle exec rails g active_record:turntable:install
generator creates Rails.root/config/turntable.yml
== ActiveRecord::Turntable terms
* Shard
* Cluster - a cluster of shards.
* Master - default ActiveRecord::Base's connection
* Sequencer - a module that creates global sequence number
== Example
=== Example Databases Structure
one main database(default ActiveRecord::Base connection) and
three user databases sharded by user_id.
+-------+
| App |
+-------+
|
+---------+-----------+---------+---------+
| | | | |
+--------+ +---------+ +-------+ +-------+ +-------+
| Master | |Sequencer| |UserDB1| |UserDB2| |UserDB3|
+--------+ +---------+ +-------+ +-------+ +-------+
=== Example Configuration
Edit turntable.yml and database.yml. See below example config.
example turntable.yml:
development:
clusters:
user_cluster: # <-- cluster name
algorithm: range # <-- only range or range_bsearch sharding is supported currently.
seq:
connection: user_seq_1
shards:
- connection: user_shard_1 # <-- shard name
less_than: 100 # <-- shard range(like mysql partitioning)
- connection: user_shard_2
less_than: 200
- connection: user_shard_3
less_than: 2000000000
database.yml:
connection_spec: &spec
adapter: mysql2
encoding: utf8
reconnect: false
pool: 5
username: root
password: root
socket: /tmp/mysql.sock
development:
<<: *spec
database: sample_app_development
seq: # <-- sequence definition
user_seq_1:
<<: *spec
database: sample_app_user_seq_development
shards: # <-- shards definition
user_shard_1:
<<: *spec
database: sample_app_user1_development
user_shard_2:
<<: *spec
database: sample_app_user2_development
user_shard_3:
<<: *spec
database: sample_app_user3_development
=== Migration Example
Generate a model:
bundle exec rails g model user name:string
Edit migration file:
class CreateUsers < ActiveRecord::Migration
clusters :user_cluster # <-- the cluster that executes migration
def change
create_table :users do |t|
t.string :name
t.timestamps
end
create_sequence_for(:users)
end
end
then please execute rake tasks:
bundle exec rake db:create
bundle exec rake db:migrate
Those rake tasks would be executed to shards too.
=== Model Example
add turntable [shard_key_name] to the model class:
class User < ActiveRecord::Base
turntable :user_cluster, :id
sequencer
has_one :status
end
class Status < ActiveRecord::Base
turntable :user_cluster, :user_id
sequencer
belongs_to :user
end
== Usage
=== Create
> User.create(name: "hoge")
(0.0ms) [Shard: user_seq_1] BEGIN
(0.3ms) [Shard: user_seq_1] UPDATE `users_id_seq` SET id=LAST_INSERT_ID(id+1)
(0.8ms) [Shard: user_seq_1] COMMIT
(0.1ms) [Shard: user_seq_1] SELECT LAST_INSERT_ID()
(0.1ms) [Shard: user_shard_1] BEGIN
[ActiveRecord::Turntable] Sending method: insert, sql: #, shards: ["user_shard_1"]
SQL (0.8ms) [Shard: user_shard_1] INSERT INTO `users` (`created_at`, `id`, `name`, `updated_at`) VALUES ('2012-04-10 03:59:42', 2, 'hoge', '2012-04-10 03:59:42')
(0.4ms) [Shard: user_shard_1] COMMIT
=> #
=== Retrieve
> user = User.find(2)
[ActiveRecord::Turntable] Sending method: select_all, sql: #, shards: ["user_shard_1"]
User Load (0.3ms) [Shard: user_shard_1] SELECT `users`.* FROM `users` WHERE `users`.`id` = 2 LIMIT 1
=> #
=== Update
> user.update_attributes(name: "hogefoo")
(0.1ms) [Shard: user_shard_1] BEGIN
[ActiveRecord::Turntable] Sending method: update, sql: UPDATE `users` SET `name` = 'hogefoo', `updated_at` = '2012-04-10 04:07:52' WHERE `users`.`id` = 2, shards: ["user_shard_1"]
(0.3ms) [Shard: user_shard_1] UPDATE `users` SET `name` = 'hogefoo', `updated_at` = '2012-04-10 04:07:52' WHERE `users`.`id` = 2
(0.8ms) [Shard: user_shard_1] COMMIT
=> true
=== Delete
> user.destroy
(0.2ms) [Shard: user_shard_1] BEGIN
[ActiveRecord::Turntable] Sending method: delete, sql: #, shards: ["user_shard_1"]
SQL (0.3ms) [Shard: user_shard_1] DELETE FROM `users` WHERE `users`.`id` = 2
(1.7ms) [Shard: user_shard_1] COMMIT
=> #
=== Count
> User.count
[ActiveRecord::Turntable] Sending method: select_value, sql: #, shards: ["user_shard_1", "user_shard_2", "user_shard_3"]
(0.8ms) [Shard: user_shard_1] SELECT COUNT(*) FROM `users`
(0.3ms) [Shard: user_shard_2] SELECT COUNT(*) FROM `users`
(0.2ms) [Shard: user_shard_3] SELECT COUNT(*) FROM `users`
=> 1
=== Sequencer
Sequencer provides generating global IDs.
Add below to the migration:
create_sequence_for(:users) # <-- this line creates sequence table
This will creates sequence table.
Next, add sequencer to the model:
class User < ActiveRecord::Base
turntable :id
sequencer # <-- this line enables sequencer module
has_one :status
end
=== Transaction
> user = User.find(2)
> user3 = User.create(name: "hoge3")
> User.shards_transaction([user, user3]) do
> user.name = "hogehoge"
> user3.name = "hogehoge3"
> user.save!
> user3.save!
> end
=== Send query to a specific shard.
use with_shard method:
AR::Base.connection.with_shard(shard1) do
# something queries to shard1
end
To access shard objects, use below:
* AR::Base.connection.shards # \\{shard_name => shard_obj,....}
* AR::Base#turntable_shard
* AR::Base.connection.select_shard(shard_key_value) #=> shard
=== Send query to all shards
use with_all method:
User.connection.with_all do
User.order("created_at DESC").limit(3).all
end
=== Migration
If you specify cluster or shard, migration will be executed to the cluster(or shard) and master database.
to specify cluster:
class CreateUsers < ActiveRecord::Migration
clusters :user_cluster
....
end
to specify shard:
class CreateUsers < ActiveRecord::Migration
shards :user_shard_01
....
end
if you would like to execute a migration to all databases in the environments, use :all symbol:
class CreateUsers < ActiveRecord::Migration
clusters :all
....
end
== Connection Management
Rails's ConnectionManagement middleware keeps ActiveRecord's connection during the process is alive, but Turntable keeps more connections.
This may cause flooding max connections on your database. So, we made a middleware that disconnects on each request.
To use turntable's ConnectionManagement middleware, add below line to your initializer.
app.middleware.swap ActiveRecord::ConnectionAdapters::ConnectionManagement, ActiveRecord::Turntable::Rack::ConnectionManagement
== Inability
* Queries includes "ORDER BY", "GROUP BY" and "LIMIT" clauses cannot be distributed.
* "has many through" and "habtm" relationships causes wrong results. ex) User-Friend-User relation
== Thanks to
ConnectionProxy, Distributed Migration implementation is inspired by Octopus and DataFabric.
== License
activerecord-turntable is released under the MIT license:
Copyright (c) 2012 Drecom Co.,Ltd.
Permission is hereby granted, free of charge, to any person obtaining
a copy of this software and associated documentation files (the
"Software"), to deal in the Software without restriction, including
without limitation the rights to use, copy, modify, merge, publish,
distribute, sublicense, and/or sell copies of the Software, and to
permit persons to whom the Software is furnished to do so, subject to
the following conditions:
The above copyright notice and this permission notice shall be
included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.