Open Source MDS

The Talk

You can watch a recording of the talk on Youtube

This stack is built on a combination of tools including:

• Dagster
• DuckDB
• dbt
• dbt-duckdbt
• Evidence
• Sling
• Steampipe

Requirements

You will need Python installed. This was all tested on Python 3.10.12 From a virtual environment, run

pip install -e .'[dev]'

Most of the dependencies will be installed through Python.

For Evidence.dev, you will need nodejs installed

Install Sling for getting data from Postgres.

# On Mac, view the website for other platforms
brew install slingdata-io/sling/sling

Steampipe is a separate requirement for the Mastodon API, and can be installed by following the instructions here

On Apple, run:

brew install turbot/tap/steampipe
steampipe plugin install turbot/mastodon

For the Mastodon API, create an Access Token. I used the birds.town instance.

Update ~/.steampipe/config/mastodon.spc with your token and instance:

connection "mastodon" {
    plugin = "mastodon"
    server = "https://birds.town"
    access_token = "abcd12345supersecretpassword"
    max_toots = -1
}

And run

dagster dev

Load up dagster at http://localhost:3000/asset-groups/

And click Materialize all to run the end-to-end pipeline.

Visualization

Evidence.dev is used for visualization.

First, go the dbt_project folder

cd dbt_project

npm --prefix ./reports install
npm --prefix ./reports run dev -- --port 4000