Postgres to Iceberg in 13 minutes: How Supermetal compares to Flink, Kafka Connect, and Spark

The real story here isn't the 13-minute benchmark—it's what that speed reveals about where CDC tooling is heading. I've spent years wrestling with Postgres replication pipelines, and the emergence of purpose-built connectors like Supermetal's Iceberg sink signals a fundamental shift: the data platform abstraction layer is moving rightward, away from orchestration-heavy Kafka and Flink topologies toward direct source-to-lakehouse paths. This matters because it directly impacts team composition and operational burden. Instead of maintaining Kafka clusters, schema registries, and complex Flink jobs just to land CDC data into Iceberg, smaller teams can now achieve comparable throughput with significantly fewer moving parts. The architectural implication is stark—we're seeing convergence toward simpler, less distributed systems that still meet SLA requirements. My recommendation: audit your current CDC infrastructure honestly. If you're running Kafka Connect or Flink primarily as a bridge to Iceberg, explore whether newer tools reduce your operational footprint without sacrificing reliability or cost.