Skip to content

Latest commit

 

History

History
293 lines (203 loc) · 7.48 KB

README.md

File metadata and controls

293 lines (203 loc) · 7.48 KB

TimescaleDB extension for Rails Gem Version Actions Status

timescaledb-rails extends ActiveRecord PostgreSQL adapter and provides features from TimescaleDB. It provides support for hypertables and other features added by TimescaleDB PostgreSQL extension.

Installation

Install timescaledb-rails from RubyGems:

$ gem install timescaledb-rails

Or include it in your project's Gemfile with Bundler:

gem 'timescaledb-rails', '~> 0.1'

Usage

Migrations

Create a hypertable from a PostgreSQL table

class CreateEvent < ActiveRecord::Migration[7.0]
  def change
    create_table :events, id: false do |t|
      t.string :name, null: false
      t.time :occurred_at, null: false

      t.timestamps
    end

    create_hypertable :events, :created_at, chunk_time_interval: '2 days'
  end
end

Create a hypertable without a PostgreSQL table

class CreatePayloadHypertable < ActiveRecord::Migration[7.0]
  def change
    create_hypertable :payloads, :created_at, chunk_time_interval: '5 days' do |t|
      t.string :ip, null: false

      t.timestamps
    end
  end
end

Add hypertable compression policy

class AddEventCompressionPolicy < ActiveRecord::Migration[7.0]
  def up
    enable_hypertable_compression :events, segment_by: :name, order_by: 'occurred_at DESC'

    add_hypertable_compression_policy :events, 20.days
  end

  def down
    remove_hypertable_compression_policy :events

    disable_hypertable_compression :events
  end
end

Add hypertable retention policy

class AddEventRetentionPolicy < ActiveRecord::Migration[7.0]
  def up
    add_hypertable_retention_policy :events, 1.year
  end

  def down
    remove_hypertable_retention_policy :events
  end
end

Add hypertable reorder policy

class AddEventReorderPolicy < ActiveRecord::Migration[7.0]
  def up
    add_hypertable_reorder_policy :events, :index_events_on_created_at_and_name
  end

  def down
    remove_hypertable_reorder_policy :events
  end
end

Create continuous aggregate

class CreateTemperatureEventAggregate < ActiveRecord::Migration[7.0]
  disable_ddl_transaction!

  def up
    create_continuous_aggregate(
      :temperature_events,
      Event.time_bucket(1.day).avg(:value).temperature.to_sql
    )

    add_continuous_aggregate_policy(:temperature_events, 1.month, 1.day, 1.hour)
  end

  def down
    drop_continuous_aggregate(:temperature_events)

    remove_continuous_aggregate_policy(:temperature_events)
  end
end

Reversible Migrations:

Above examples implement up/down methods to better document all the different APIs. Feel free to use change method, timescaledb-rails defines all the reverse calls for each API method so Active Record can automatically figure out how to reverse your migration.

Models

If one of your models need TimescaleDB support, just include Timescaledb::Rails::Model

class Payload < ActiveRecord::Base
  include Timescaledb::Rails::Model

  self.primary_key = 'id'
end

When hypertable belongs to a non default schema, don't forget to override table_name

class Event < ActiveRecord::Base
  include Timescaledb::Rails::Model

  self.table_name = 'tdb.events'
end

Using .find is not recommended, to achieve more performant results, use these other find methods

# When you know the exact time value
Payload.find_at_time(111, Time.new(2022, 01, 01, 10, 15, 30))

# If you know that the record occurred after a given time
Payload.find_after(222, 11.days.ago)

# Lastly, if you want to scope the search by a time range
Payload.find_between(333, 1.week.ago, 1.day.ago)

If you need to query data for a specific time period, Timescaledb::Rails::Model includes useful scopes

# If you want to get all records from last year
Event.last_year #=> [#<Event name...>, ...]

# Or if you want to get records from this year
Event.this_year #=> [#<Event name...>, ...]

# Or even getting records from today
Event.today #=> [#<Event name...>, ...]

Here the list of all available scopes

  • last_year
  • last_month
  • last_week
  • this_year
  • this_month
  • this_week
  • yesterday
  • today

If you still need to query data by other time periods, take a look at these other scopes

# If you want to get all records that occurred in the last 30 minutes
Event.after(30.minutes.ago) #=> [#<Event name...>, ...]

# If you want to get records that occurred in the last 4 days, excluding today
Event.between(4.days.ago, 1.day.ago) #=> [#<Event name...>, ...]

# If you want to get records that occurred at a specific time
Event.at_time(Time.new(2023, 01, 04, 10, 20, 30)) #=> [#<Event name...>, ...]

If you need information about your hypertable, use the following helper methods to get useful information

# Hypertable metadata
Event.hypertable #=> #<Timescaledb::Rails::Hypertable ...>

# Hypertable chunks metadata
Event.hypertable_chunks #=> [#<Timescaledb::Rails::Chunk ...>, ...]

# Hypertable jobs, it includes jobs like compression, retention or reorder policies, etc.
Event.hypertable_jobs #=> [#<Timescaledb::Rails::Job ...>, ...]

# Hypertable dimensions, like time or space dimensions
Event.hypertable_dimensions #=> [#<Timescaledb::Rails::Dimension ...>, ...]

# Hypertable compression settings
Event.hypertable_compression_settings #=> [#<Timescaledb::Rails::CompressionSetting ...>, ...]

If you need to compress or decompress a specific chunk

chunk = Event.hypertable_chunks.first

chunk.compress! unless chunk.is_compressed?

chunk.decompress! if chunk.is_compressed?

If you need to reorder a specific chunk

chunk = Event.hypertable_chunks.first

# If an index is not specified, it will use the one from the reorder policy
# In case there is no reorder policy index it will raise an ArgumentError
chunk.reorder!

# If an index is specified it will use that index
chunk.reorder!(index)

If you need to manually refresh a continuous aggregate

aggregate = Event.hypertable.continuous_aggregates.first

aggregate.refresh!(5.days.ago, 1.day.ago)

Hyperfunctions

Time bucket

You can call the time bucket function with an interval (note that leaving the target column blank will use the default time column of the hypertable)

Event.time_bucket(1.day)

Event.time_bucket('1 day')

Event.time_bucket(1.day, :created_at)

Event.time_bucket(1.day, 'occurred_at')

You may add aggregation like so:

Event.time_bucket(1.day).avg(:column)
Event.time_bucket(1.day).sum(:column)
Event.time_bucket(1.day).min(:column)
Event.time_bucket(1.day).max(:column)
Event.time_bucket(1.day).count

Contributing

Please read CONTRIBUTING.md for details on our code of conduct, and the process for submitting pull requests.

Supported Ruby/Rails versions

Supported Ruby/Rails versions are listed in .github/workflows/ci.yaml

License

Released under the MIT License. See the LICENSE file for further details.

About Crunchloop

crunchloop

timescaledb-rails is supported with ❤️ by Crunchloop. We strongly believe in giving back 🚀. Let's work together Get in touch.