Skip to content

craig-day/ecto_sharding

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

63 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

EctoSharding

Build Status Coverage Status

A simple sharding library for Ecto.

Documentation can be viewed on hexdocs.

Installation

defp deps do
  [
    {:ecto_sharding, "~> 0.0.8"}
  ]
end

Usage

Configuration

  1. Configure your app's Ecto.Repo

    This configuration will be used when a not-sharded schema is being queried.

    config :my_app, MyApp.Repo,
      adapter: Ecto.Adapters.MySQL,
      username: System.get_env("MYSQL_USERNAME"),
      password: System.get_env("MYSQL_PASSWORD"),
      database: System.get_env("MYSQL_DATABASE"),
      hostname: System.get_env("MYSQL_HOST"),
      pool_size: 15
  2. Configure EctoSharding

    otp_app needs to be set so EctoSharding knows how to builds the repos for each shard.

    config :ecto_sharding, EctoSharding,
      otp_app: :my_app

    If you know your shard information at compile time, you can also add that.

    Note: Be sure to set the priv key for the shard repos to something different than the main db. Ecto uses the priv key to determine where to put the migrations folder and schema files for a Repo. Defining this for the shards Repos avoids mixing the main db and shard db files in the same directory. In the example below, the shard db migrations will be in priv/shards/migrations, and the main db files will be in priv/repo/migrations (the Ecto default location when priv is not specified).

    config :ecto_sharding, EctoSharding,
      otp_app: :my_app,
      shards: %{
        "shard_1" => [
          adapter: Ecto.Adapters.MySQL,
          username: System.get_env("MYSQL_USERNAME"),
          password: System.get_env("MYSQL_PASSWORD"),
          database: "my_db_shard_1",
          hostname: "10.0.0.1",
          pool_size: 15,
          priv: "priv/shards"
        ],
        "shard_2" => [
          adapter: Ecto.Adapters.MySQL,
          username: System.get_env("MYSQL_USERNAME"),
          password: System.get_env("MYSQL_PASSWORD"),
          database: "my_db_shard_2",
          hostname: "10.0.0.2",
          pool_size: 15,
          priv: "priv/shards"
        ]
      }
  3. Start EctoSharding as a supervised process

    When the supervisor starts up, it will expect all of its configuration to be there, so make sure any config is set before start_link is called.

    supervisor(EctoSharding, [])
  4. use EctoSharding.Repo instead of Ecto.Repo

    In any repos you have defined in your app, use EctoSharding.Repo instead:

    use EctoSharding.Repo, otp_app: :my_app
  5. use EctoSharding.Schema instead of Ecto.Schema

    The schema is where most of the magic happens. This is where you can declare a particular schema as sharded or not. Schemas will default to sharded.

    A sharded schema

    defmodule MyApp.User do
      use EctoSharding.Schema, sharded: true # default
    
      schema "users" do
        field :name, :string
        # ...
      end
    end

    A not-sharded schema

    defmodule MyApp.Account do
      use EctoSharding.Schema, sharded: false
    
      schema "accounts" do
        field :name, :string
        # ...
      end
    end

Querying

Setting the current shard

EctoSharding uses a shard registry, backed by GenServer to store information about which shard we currently want to use and how to talk to that shard. This means that you need to set the current shard in your application before you can issue a query to the sharded database.

Setting the shard is as simple as

EctoSharding.current_shard("shard_1")

Take a plug based web application for example. This is what a plug that sets the current shard might look like:

defmodule MyApp.ShardContext do
  @behaviour Plug

  import Plug.Conn

  def init(opts), do: opts

  def call(conn, _) do
    account = conn.assigns.account
    EctoSharding.current_shard(account.shard_id)
    conn
  end
end

Executing a query

Once the current shard has been set, we can query our repo just like normal and it will take care of using the correct repo for the schema involved in the query.

import Ecto.Query

MyApp.User
|> where(name: "Jane Doe")
|> limit(1)
|> MyApp.Repo

About

Database sharding support for Elixir Ecto

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages