We've covered streaming processing (Day 1), reliability (Day 2), and CDC replication (Day 3) as individual technologies. Today we step back to see how to weave them into one system. In data engineering, "which service?" matters less than "how do we arrange the pieces?" Same components become robust pipelines or brittle spaghetti based on topology.
Today we cover two axes: first, processing architecture handling batch and stream together (Lambda vs Kappa); second, loose coupling via event-driven ingestion (EventBridge, SQS, SNS).
Lambda Architecture (unrelated to AWS Lambda service) is a pattern sending same data down two paths.