A single Postgres operational database has two tables of interest

A medium Pipeline Design interview practice problem on DataDriven. Write and execute real pipeline design code with instant grading.

Domain

Pipeline Design

Difficulty

medium

Problem

A single Postgres operational database has two tables of interest. customers (~2M rows, slowly changing) needs hourly freshness; orders (~500M rows, +10M/day) needs five-minute freshness. The destination is Snowflake. The section's three legitimate strategies: full load + incremental pull (cheapest; no deletes on the incremental side), trigger-based CDC (captures deletes; doubles writes), and log-based CDC with Debezium (low overhead; needs WAL access and slot management). Pick the CDC strategy by adding a strategy-marker node downstream of the database whose name states the chosen strategy per table and the trade-off that drove it (scale, deletes, latency, or team capacity).

Practice This Problem

Solve this Pipeline Design problem with real code execution. DataDriven runs your solution and grades it automatically.