A customer's Delta table has 4 million small files and every query crawls. What happened, and how do you fix it?

FDEInterviews · 2026-06-14 · 1 min read

Summary

This article uses a realistic scenario—a customer's Delta table with 4 million small files making every query slow—to show how candidates should handle the small-files problem in Forward Deployed Engineer interviews. The author calls it the most common self-inflicted lakehouse issue seen in the field, with causes that reduce to the same three patterns. A complete answer requires both a fix and a prevention plan, which most candidates forget. Interviewers reward candidates who first explain where the obvious approach breaks down, trace the failure backward through the pipeline, and identify the one metric the customer's executive sponsor cares about before proposing a solution.

Key points

Scenario: a Delta table with 4 million small files slows all queries.
Small files are the top self-inflicted lakehouse problem FDEs encounter.
Root causes typically fall into three recurring patterns.
A strong answer includes prevention, not just the immediate fix.
Lead with where the obvious approach fails, then trace the pipeline backward.

Written by AI from the full original article.

Full text

A customer's Delta table has 4 million small files and every query crawls. What happened, and how do you fix it?

The small-files problem is the #1 self-inflicted lakehouse wound FDEs find in the field. The causes are always the same three things, and the fix has a prevention half most candidates forget.

Updated Aug 2026 · Grounded in real Forward Deployed Engineer interview loops and written to a senior-engineer editorial bar.

The small-files problem is the #1 self-inflicted lakehouse wound FDEs find in the field. The causes are always the same three things, and the fix has a prevention half most candidates forget.

Lead with where the obvious approach breaks, because that is the judgment they are screening for — most candidates jump straight to the happy path and lose the room.

Then walk the failure back through the pipeline in order, naming the one metric the customer's exec sponsor actually cares about before you propose the fix.

Summary

Key points

Full text

Related reading