Checkpointing: Cutting the Lineage

Checkpointing is the deliberate act of truncating a lineage. When you checkpoint a DataFrame, Spark computes it and writes the result to reliable storage, then discards the lineage behind it. From that point on, the checkpointed data is a new starting point with no history: if a downstream partition is lost, Spark recovers it by reading the checkpoint, not by replaying the entire chain that produced it. You have traded the recompute cost for a one-time write cost. This matters most for the two expensive cases from the last section. For a very long iterative lineage, checkpointing every so often caps how far back any recovery has to replay: instead of going to the original source, it goes to the most recent checkpoint. For a pipeline that has already done expensive wide work, checkpointing

About This Interactive Section

This section is part of the Lineage as Fault Tolerance lesson on DataDriven, a free data engineering interview prep platform. Each section includes explanations, worked examples, and hands-on code challenges that execute in real time. SQL queries run against a live PostgreSQL database. Python runs in a sandboxed Docker container. Data modeling problems validate against interactive schema canvases. All content is framed around what data engineering interviewers actually test at companies like Meta, Google, Amazon, Netflix, Stripe, and Databricks.

How DataDriven Lessons Work

DataDriven combines four interview rounds (SQL, Python, Data Modeling, Pipeline Architecture) with adaptive difficulty and spaced repetition. Easy problems get harder as you improve. Weak concepts resurface until you master them. Your readiness score tracks progress across every topic interviewers test. Every lesson section ends with problems you solve by writing and running real code, not by picking multiple-choice answers.