Great overview of the shift from manual infrastructure to managed services. You mention developers previously had to manually optimize performance—could you elaborate on what specific metrics or levers (like latency vs. throughput) we should still monitor closely within Bedrock's managed environment?