tag: observability

AWS just dropped native Application Load Balancer log support in CloudWatch Logs, and it's a game-changer for debugging network traffic. You can now query ALB access, connection, and health check logs directly with CloudWatch Logs Insights, set up automatic telemetry rules for consistent monitoring across your org, and even stream logs to Data Firehose or S3 in Parquet format. Real-time traffic analysis via Live Tail plus metric filters for alerting? Chef's kiss. Configure it through the AWS Console, CLI, or SDKs—available everywhere CloudWatch runs.

Amazon Bedrock AgentCore just made debugging way less painful—all agent traces, prompts, and logs now land in a single CloudWatch log group instead of scattered across multiple destinations. Previously, you'd hunt through different log groups to piece together what your agent actually did; now everything's unified per-agent with proper IAM and encryption controls. New agents get this by default starting July 20, 2026, while existing ones just need to flip `UNIFIED_TRACES_DESTINATION_ENABLED=true` and upgrade ADOT to 0.17.1+. For multi-agent systems, this is a game-changer—end-to-end debugging just became straightforward.