Production support requires more engineers than actual product development
Logs, DAGs, metrics, and datasets often fail quietly, forcing teams to stitch together schedulers, observability, and enrichment tools just to find the source. Each investigation pulls multiple engineers into hours of manual work. Small changes can cascade across the stack, turning minor issues into expensive outages




