The Recompute Cost: When Lineage Gets Expensive

Recovery is cheap when the lineage is short and narrow, because replaying it touches little data and moves none across the network. It gets expensive in two ways, and once you can recognise them you can engineer for the failure instead of trusting fault tolerance blindly. The first is length. Every transformation you chain adds a step to the lineage of the partitions it produces. A pipeline with hundreds of transformations, common in iterative algorithms that loop and refine, builds a very long lineage, and recovering a lost partition near the end means replaying that entire long recipe from the source. The recovery is correct, but it can take almost as long as the original computation for that partition. The second, and worse, is width. A narrow dependency replays cheaply: one output part

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.