Six Main Components of a Data Pipeline

How does a data engineer develop a data pipeline?

Naser Tamimi
Published in
4 min readNov 4, 2022


Photo by Simone Hutsch on Unsplash

Developing a new data pipeline is a long process. It usually starts with the business justification, defining metrics, researching available data and tables, and finally, developing and launching the pipeline. In this article, I specifically focus on the development stage. Assume, at this step; you know the business requirements, worked with other stakeholders to define your metrics, made sure that the required data is available on other tables, and you are finally ready to start developing your pipeline. Here are six components that you should have in your new data pipeline.

1. General Parameters

Your pipeline has some general settings which should be set at the beginning. Parameters such as schedule interval, start date, credential parameters, point of contact if it fails, and whether it should run if the previous partitions failed are just a few examples of the parameters that should be set at the beginning of developing a data pipeline.

Try to understand the consequences of setting each of these parameters. For example, suppose your pipeline aggregates a metric from the past partitions. In that case, you should set an appropriate parameter to fail…