Handle Late Data in Streaming

HardPipelines00:00

Practice interviewer

Your interviewer

In session

Interviewer

Welcome to your interview.

The question is on your right: Handle Late Data in Streaming. Take a moment with it first.

Talk your thinking through with me if you like - when you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes). Discussion and graded submissions share your five interviewer interactions, so spend them well.

You need to log in / sign up to chat or submit.

Problem

You are designing a streaming pipeline where some events arrive after their event time. Those late records can invalidate previously computed windows, aggregates, or fact tables. The goal is to keep downstream data correct while limiting unnecessary recomputation.

What this tests

Event-time processing and watermark design
Idempotent writes and deduplication
Repair strategy for very late records
Data quality controls for corrected outputs

Key design signals

Common trade-off·Longer watermark vs lower latencyRecovery pattern·Targeted backfill instead of full replayPrimary challenge·Correctness under out-of-order arrival