CHAPTER 3Data Pipelines
You just launched a high-visibility feature meant to improve retention. As the lead engineer, you’re in charge of the post-launch debrief. It’s your job to walk stakeholders through the data and show how the feature performed.
You open the dashboards.
They’re blank.
No events. No charts. No usage insights.
After some digging, you realize the ingestion job that collects telemetry and sends it to your dashboards has failed. Worse, there was no alerting in place to let you know.
The launch happened, but the data didn’t follow.
This chapter is about that data pipeline—how telemetry moves through systems, what can go wrong, and how to design pipelines that are reliable, observable, and ready to support growth.
What Is a Data Pipeline and Why Does It Matter?
When you think about a pipeline, what comes to mind?
You may think of a long tunnel or tube that carries water or oil. If you’re a hardware engineer, you might think of the five-stage processor pipeline: fetch, decode, execute, load, and store. If you work in a laundromat, you might imagine the dry-cleaning process: collect clothes, wash, dry, iron, and fold.
Whatever comes to mind, you’re not wrong. Each of these is a step-by-step system that moves something from one stage to another.
That’s exactly what a data pipeline is.
In this context, a data pipeline is a set of systems through which telemetry flows. Telemetry includes the logs, metrics, and events generated by your application. A pipeline starts ...
Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.
Read now
Unlock full access