Skip to content

Latest commit

 

History

History
490 lines (362 loc) · 15 KB

README.md

File metadata and controls

490 lines (362 loc) · 15 KB
Raw

TimescaleDB

Welcome to the TimescaleDB gem! To experiment with the code, start installing the gem:

gem install timescaledb

The tsdb CLI

When you install the gem locally, a new command line application named tsdb will be linked in your command line.

It accepts a Postgresql URI and some extra flags that can help you to get more info from your TimescaleDB server:

tsdb <uri> --stats

Where the <uri> is replaced with params from your connection like:

tsdb postgres://<user>@localhost:5432/<dbname> --stats

Or just check the stats:

tsdb "postgres://<user>@localhost:5432/timescaledb_test" --stats

These is a sample output from database example with almost no data:

{:hypertables=>
  {:count=>3,
   :uncompressed=>2,
   :chunks=>{:total=>1, :compressed=>0, :uncompressed=>1},
   :size=>{:before_compressing=>"80 KB", :after_compressing=>"0 Bytes"}},
 :continuous_aggregates=>{:count=>1},
 :jobs_stats=>[{:success=>nil, :runs=>nil, :failures=>nil}]}

To start a interactive ruby/pry console use --console: The console will dynamically create models for all hypertables that it finds in the database.

Let's consider the caggs.sql as the example of database.

psql postgres://<user>@localhost:5432/playground -f caggs.sql

Then use tsdb in the command line with the same URI and --stats:

tsdb postgres://<user>@localhost:5432/playground --stats
{:hypertables=>
  {:count=>1,
   :uncompressed=>1,
   :approximate_row_count=>{"ticks"=>352},
   :chunks=>{:total=>1, :compressed=>0, :uncompressed=>1},
   :size=>{:uncompressed=>"88 KB", :compressed=>"0 Bytes"}},
 :continuous_aggregates=>{:total=>1},
 :jobs_stats=>[{:success=>nil, :runs=>nil, :failures=>nil}]}

To have some interactive playground with the actual database using ruby, just try the same command before changing from --stats to --console:

tsdb --console

The same database from previous example, is used so the context has a hypertable named ticks and a view named ohlc_1m.

tsdb postgres://<user>@localhost:5432/playground --console
pry(Timescale)>

The tsdb CLI will automatically create ActiveRecord models for hypertables and continuous aggregates views.

Tick
=> Timescaledb::Tick(time: datetime, symbol: string, price: decimal, volume: integer)

Note that it's only created for this session and will never be cached in the library or any other place.

In this case, Tick model comes from ticks hypertable that was found in the database. It contains several extra methods inherited from acts_as_hypertable macro.

Let's start with the .hypertable method.

Tick.hypertable
=> #<Timescaledb::Hypertable:0x00007fe99c258900
 hypertable_schema: "public",
 hypertable_name: "ticks",
 owner: "jonatasdp",
 num_dimensions: 1,
 num_chunks: 1,
 compression_enabled: false,
 is_distributed: false,
 replication_factor: nil,
 data_nodes: nil,
 tablespaces: nil>

The core of the hypertables are the fragmentation of the data into chunks that are the child tables that distribute the data. You can check all chunks directly from the hypertable relation.

Tick.hypertable.chunks
unknown OID 2206: failed to recognize type of 'primary_dimension_type'. It will be treated as String.
=> [#<Timescaledb::Chunk:0x00007fe99c31b068
  hypertable_schema: "public",
  hypertable_name: "ticks",
  chunk_schema: "_timescaledb_internal",
  chunk_name: "_hyper_33_17_chunk",
  primary_dimension: "time",
  primary_dimension_type: "timestamp without time zone",
  range_start: 1999-12-30 00:00:00 +0000,
  range_end: 2000-01-06 00:00:00 +0000,
  range_start_integer: nil,
  range_end_integer: nil,
  is_compressed: false,
  chunk_tablespace: nil,
  data_nodes: nil>]

Chunks are created by partitioning a hypertable's data into one (or potentially multiple) dimensions. All hypertables are partitioned by the values belonging to a time column, which may be in timestamp, date, or various integer forms. If the time partitioning interval is one day, for example, then rows with timestamps that belong to the same day are co-located within the same chunk, while rows belonging to different days belong to different chunks. Learn more here.

Another core concept of TimescaleDB is compression. With data partitioned, it becomes very convenient to compress and decompress chunks independently.

Tick.hypertable.chunks.first.compress!
ActiveRecord::StatementInvalid: PG::FeatureNotSupported: ERROR:  compression not enabled on "ticks"
DETAIL:  It is not possible to compress chunks on a hypertable that does not have compression enabled.
HINT:  Enable compression using ALTER TABLE with the timescaledb.compress option.

As compression is not enabled, let's do it executing a plain SQL directly from the actual context. To borrow a connection, let's use the Tick object.

Tick.connection.execute("ALTER TABLE ticks SET (timescaledb.compress)") # => PG_OK

And now, it's possible to compress and decompress:

Tick.hypertable.chunks.first.compress!
Tick.hypertable.chunks.first.decompress!

Learn more about TimescaleDB compression here.

The ohlc_1m view is also available as an ActiveRecord:

Ohlc1m
=> Timescaledb::Ohlc1m(bucket: datetime, symbol: string, open: decimal, high: decimal, low: decimal, close: decimal, volume: integer)

And you can run any query as you do with regular active record queries.

Ohlc1m.order(bucket: :desc).last
=> #<Timescaledb::Ohlc1m:0x00007fe99c2c38e0
 bucket: 2000-01-01 00:00:00 UTC,
 symbol: "SYMBOL",
 open: 0.13e2,
 high: 0.3e2,
 low: 0.1e1,
 close: 0.1e2,
 volume: 27600>

Installation

Add this line to your application's Gemfile:

gem 'timescaledb'

And then execute:

$ bundle install

Or install it yourself as:

$ gem install timescaledb

Usage

Check the examples/ranking to get a Rails complete example.

You can check the all_in_one.rb example that will:

  1. Create hypertable with compression settings
  2. Insert data
  3. Run some queries
  4. Check chunk size per model
  5. Compress a chunk
  6. Check chunk status
  7. Decompress a chunk

Toolkit

Toolkit contains a lot of extra features to analyse data more deeply directly in the SQL. There are a few examples in the examples/toolkit-demo folder that can let you benchmark and see the differences between implementing the algorithm directly in Ruby or directly in SQL using the Timescaledb Toolkit extension.

For now you can benchmark and compare:

  1. volatility algorithm.
  2. lttb algorithm.

Testing

If you need some inspiration for how are you going to test your hypertables, please check the spec/spec_helper.rb for inspiration.

Migrations

Create table is now with the hypertable keyword allowing to pass a few options to the function call while also using create_table method:

create_table with :hypertable

hypertable_options = {
  time_column: 'created_at',
  chunk_time_interval: '1 min',
  compress_segmentby: 'identifier',
  compression_interval: '7 days'
}

create_table(:events, id: false, hypertable: hypertable_options) do |t|
  t.string :identifier, null: false
  t.jsonb :payload
  t.timestamps
end

create_continuous_aggregate

This example shows a ticks table grouping ticks as OHLCV histograms for every minute.

hypertable_options = {
  time_column: 'created_at',
  chunk_time_interval: '1 min',
  compress_segmentby: 'symbol',
  compress_orderby: 'created_at',
  compression_interval: '7 days'
}
create_table :ticks, hypertable: hypertable_options, id: false do |t|
  t.string :symbol
  t.decimal :price
  t.integer :volume
  t.timestamps
end
Tick = Class.new(ActiveRecord::Base) do
  self.table_name = 'ticks'
  self.primary_key = 'symbol'
  acts_as_hypertable
end

query = Tick.select(<<~QUERY)
  time_bucket('1m', created_at) as time,
  symbol,
  FIRST(price, created_at) as open,
  MAX(price) as high,
  MIN(price) as low,
  LAST(price, created_at) as close,
  SUM(volume) as volume").group("1,2")
QUERY

options = {
  with_data: false,
  refresh_policies: {
    start_offset: "INTERVAL '1 month'",
    end_offset: "INTERVAL '1 minute'",
    schedule_interval: "INTERVAL '1 minute'"
  }
}

create_continuous_aggregate('ohlc_1m', query, **options)

Scenic integration

The Scenic gem is an easy way to manage database view definitions for a Rails application. TimescaleDB's continuous aggregates are more complex than regular PostgreSQL views, and the schema dumper included with Scenic can't dump a complete definition.

This gem automatically configures Scenic to use a Timescaledb::Scenic::Adapter which will correctly handle schema dumping.

Enable ActsAsHypertable

You can declare a Rails model as a Hypertable by invoking the acts_as_hypertable macro. This macro extends your existing model with timescaledb-related functionality. model:

class Event < ActiveRecord::Base
  acts_as_hypertable
end

By default, ActsAsHypertable assumes a record's time_column is called created_at.

Options

If you are using a different time_column name, you can specify it as follows when invoking the acts_as_hypertable macro:

class Event < ActiveRecord::Base
  acts_as_hypertable time_column: :timestamp
end

Chunks

To get all the chunks from a model's hypertable, you can use .chunks.

Event.chunks # => [#<Timescaledb::Chunk>, ...]

Hypertable metadata

To get the models' hypertable metadata, you can use .hypertable.

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

To get hypertable metadata for all hypertables: Timescaledb.hypertables.

Compression Settings

Compression settings are accessible through the hypertable.

Event.hypertable.compression_settings # => [#<Timescaledb::CompressionSettings>, ...]

To get compression settings for all hypertables: Timescaledb.compression_settings.

Scopes

The acts_as_hypertable macro can be very useful to generate some extra scopes for you. Example of a weather condition:

class Condition < ActiveRecord::Base
  acts_as_hypertable time_column: "time"
end

Through the ActsAsHypertable on the model, a few scopes are created based on the time_column argument:

Scope name What they return
Model.previous_month Records created in the previous month
Model.previous_week Records created in the previous week
Model.this_month Records created this month
Model.this_week Records created this week
Model.yesterday Records created yesterday
Model.today Records created today
Model.last_hour Records created in the last hour

All time-related scopes respect your application's timezone.

When you enable ActsAsTimeVector on your model, we include a couple default scopes. They are:

class Condition < ActiveRecord::Base
  acts_as_time_vector time_column: "time",
    value_column: "temperature",
    segment_by: "device_id"
end

RSpec Hooks

In case you want to use TimescaleDB on a Rails environment, you may have some issues as the schema dump used for tests does not consider hypertables metadata.

As a work around, you can dynamically create the hypertables yourself for testing environments using the following hook which you can define in spec/rspec_helper.rb:

config.before(:suite) do
  hypertable_models = ActiveRecord::Base.descendants.select(&:acts_as_hypertable?)

  hypertable_models.each do |klass|
    table_name = klass.table_name
    time_column = klass.hypertable_options[:time_column]

    if klass.try(:hypertable).present?
      ApplicationRecord.logger.info "hypertable already created for '#{table_name}', skipping."
      next
    end

    ApplicationRecord.connection.execute <<~SQL
      SELECT create_hypertable('#{table_name}', '#{time_column.to_s}')
    SQL
  end
end

Development

After checking out the repo, run bin/setup to install the development dependencies. Then, bundle exec rake test:setup to setup the test database and tables. Finally, run bundle exec rake to run the tests.

You can also run tsdb for an interactive prompt that will allow you to experiment.

To install this gem onto your local machine, run bundle exec rake install. To release a new version, update the version number in version.rb, and then run bundle exec rake release, which will create a git tag for the version, push git commits and tags, and push the .gem file to rubygems.org.

You can create a .env file locally to run tests locally. Make sure to put your own credentials there!

PG_URI_TEST="postgres://<user>@localhost:5432/<dbname>"

You can put some postgres URI directly as a parameter of tsdb. Here is an example from the console:

tsdb "postgres://<user>@localhost:5432/timescaledb_test"

More resources

This library was started on twitch.tv/timescaledb. You can watch all episodes here:

  1. Wrapping Functions to Ruby Helpers.
  2. Extending ActiveRecord with Timescale Helpers.
  3. Setup Hypertables for Rails testing environment.
  4. Packing the code to this repository.
  5. the code to this repository.
  6. Working with Timescale continuous aggregates.
  7. Creating the command-line application in Ruby to explore the Timescale API.

TODO

Here is a list of functions that would be great to have:

  • Add data nodes support

Contributing

Bug reports and pull requests are welcome on GitHub at https://github.com/jonatas/timescaledb. This project is intended to be a safe, welcoming space for collaboration, and contributors are expected to adhere to the code of conduct.

License

The gem is available as open source under the terms of the MIT License.

Code of Conduct

Everyone interacting in the Timescale project's codebases, issue trackers, chat rooms and mailing lists is expected to follow the code of conduct.