Airflow VS dagster

💡 You can read more here.

Its a bit overkill but i use Airflow with local executor.

To learn more about it, I built a Data Pipeline that uses Apache Airflow to pull Elon Musk tweets using the Twitter API and store the result in a CSV stored in a MinIO (OSS alternative to AWS s3) Object Storage bucket.

Airflow

I am currently having a problem with trying to enable Azure OAuth to authenticate into our airflow instance. I have posted in countless other places trying to get answers so this is my next place I am trying. Here is the link to the discussion I posted within the airflow repo: https://github.com/apache/airflow/discussions/28098 but I will also do the liberty of posting it here as well. If anybody has any knowledge or can help I would greatly appreciate it as I have been dealing with this for over a month with no answers.

Airflow is really popular, started at Airbnb. Pros: huge community, super mature. Cons: generic workflow orchestration, not the best for handling only data, hard to scale and maintain.

Airflow might also be a good option for you. Essentially DAGs of cronjobs. We like it a lot.

$ helm upgrade --install airflow apache-airflow/airflow --namespace airflow --create-namespace Release "airflow" does not exist. Installing it now. NAME: airflow LAST DEPLOYED: Sun Nov 6 02:06:55 2022 NAMESPACE: airflow STATUS: deployed REVISION: 1 TEST SUITE: None NOTES: Thank you for installing Apache Airflow 2.4.1! Your release is named airflow. You can now access your dashboard(s) by executing the following command(s) and visiting the corresponding port at localhost in your browser: Airflow Webserver: kubectl port-forward svc/airflow-webserver 8080:8080 --namespace airflow Default Webserver (Airflow UI) Login credentials: username: admin password: admin Default Postgres connection credentials: username: postgres password: postgres port: 5432 You can get Fernet Key value by running the following: echo Fernet Key: $(kubectl get secret --namespace airflow airflow-fernet-key -o jsonpath="{.data.fernet-key}" | base64 --decode) ########################################################### # WARNING: You should set a static webserver secret key # ########################################################### You are using a dynamically generated webserver secret key, which can lead to unnecessary restarts of your Airflow components. Information on how to set a static webserver secret key can be found here: https://airflow.apache.org/docs/helm-chart/stable/production-guide.html#webserver-secret-key

I gotta admit, my first thought was "Duct Size" is a weird name for a distributed work-flow tool[1].

[1] https://airflow.apache.org/

Dagster? https://dagster.io

One that I haven't seen on here yet: dagster

I wondered how it related to https://dagster.io/

You can find all current, closed, and resolved issues on the “Issues” section and explore them using filters: eg issues for dagster. Look into some of the issues and feel free to ask a question or post your idea: it’s much less toxic here (compared to SO, for example).

You could also consider Dagster, which aims to improve Apache Airflow's shortcomings. Also, take a look at MyMLOps, where you can get a quick overview of open-source orchestration tools.

Python is one of the most accessible programming l code within Python. My favorite is dagster, which forces you to write functional blocks of code with superior features—coming from a more SQL, T-SQL, and PL-SQL background. As a data engineer, I'd say you'd not expect to write perfect code; it's better to know the Big-O annotation to avoid long-running data pipelines, even if your code doesn't look the prettiest. Static types such as mypy might be another good one to know, as it will detect errors pre-runtime, which is the biggest problem of Python.

This is cool, but looks like https://github.com/dagster-io/dagster

The issue with less popular data pipeline projects is that they’re less stable in production

Papermill can also target cloud storage outputs for hosting rendered notebooks, execute notebooks from custom Python code, and even be used within distributed data pipelines like Dagster (see Dagstermill). For more information, see the papermill documentation.

There are specialized tools like DataHub (see this for columnar level reporting: https://feature-requests.datahubproject.io/roadmap/541 ) that would help. But really, in a good data platform, the orchestration layer should be aggregating metadata and giving you everything you need to trace lineage, A tool like Dagster does this well if you make full use of the Software Defined Assets capability, but that is fairly new so not so many people have embraced it yet.