Lesson 18 · Senior CDC practice

Senior CDC interview questions

High-signal CDC, Debezium, Kafka Connect, and streaming-ETL questions — with answer shapes that sound like real backend maturity instead of memorized platform slogans.

Your win: answer senior CDC and streaming-ETL questions with model + trade-off + repo-grounded consequence, not just platform definitions.

How to get the most out of this page

This page is not meant to be skimmed like a glossary. The strongest earlier lessons teach by guiding your attention. Use this one the same way: read a question, pause, answer it out loud, then compare your answer shape to the one here.

What matters is not just whether you know the term. It is whether your answer sounds like a person who understands the system under pressure.

In plain English Plain English: strong CDC answers explain what the pipeline really guarantees, what trade-off buys it, and how that shows up in a real system.

Why senior CDC answers feel different

Foundational CDC questions ask whether you know Debezium, Kafka Connect, and JDBC sinks. Senior CDC questions ask whether you can reason about those tools honestly when there is risk, ambiguity, or trade-off involved.

That is why answer shape matters so much. A good answer names the model, the important trade-off or failure risk, and the repo-grounded consequence. Without that last step, answers often sound memorized instead of owned.

The answer recipe Start with the CDC or streaming model, name the key trade-off or failure risk, then finish with the repo-grounded consequence.

Question bank

Question	Best answer shape
Why is log-based CDC better than polling here?	Because it reads committed changes from the WAL in order, captures deletes, and avoids repeated scans on the source DB.
Why isn’t at-least-once the whole safety story?	Because replay is only safe when the sink applies repeated deliveries idempotently, usually via primary-key upsert.
Why is source recreation riskier than sink recreation?	Because the source owns continuity with the WAL through the replication slot, while a sink is just a downstream applier.
What is the real value of incremental snapshots?	They let you backfill existing rows for new tables while preserving the running source connector and its slot continuity.
Why generate connectors instead of hand-maintaining them?	Because the fleet is too large and repetitive; generation plus reconcile reduces drift and makes the repo the source of truth.
What is the real gap in the generated-sink failure model?	No DLQ means poison-record recovery is weaker even though restartability and monitoring exist.
Why is schema evolution a pipeline problem?	Because one source change can affect the event schema, sink config, destination table, and warehouse logic downstream.
What should a senior review ask about a CDC design?	What exact guarantees it really provides about continuity, replay safety, schema handling, and failure behavior.
When would you choose outbox instead of plain CDC?	When you need explicit domain-event intent instead of table-replication semantics, or need a safer answer to the dual-write problem at the application boundary.
What is the strongest one-line description of hephaestus?	A GitOps-style reconcile job that converges generated connector configs into live Connect clusters while protecting risky destructive paths.

Backend use case The strongest answers mention repo facts like the three stacks, dbz_signals, generated sink whitelists, the delete-guard, no-DLQ generated sinks, ksqlDB-owned connectors, and chained CDC in lmsdwh.

Use the playbook and map together

The senior playbook compresses the judgment layer; the repo map gives you the ground truth that makes answers sound real.

→ Senior backend playbook
→ Repo CDC map

Check yourself (from memory)

Q1. A strong senior CDC answer usually includes…

That structure makes the answer sound like engineering judgment rather than glossary recall.

What are the three ingredients of a strong senior CDC answer?

recall, then click to reveal

A clear CDC or streaming model, the important trade-off or failure risk, and a repo-grounded consequence or example.

When you’re ready, don’t just reread this page — go run the mock pack out loud. Ask me if you want me to conduct it live.

Sources. Repo CDC map; curated CDC resources.