Amazon has introduced managed daemons for its Elastic Container Service (ECS) Managed Instances, a structural change for platform teams. The feature creates a separate construct for operational software like the CloudWatch Agent, Splunk forwarders, or Datadog tracers. This directly addresses a chronic pain point: updating a monitoring agent no longer forces a redeployment of application tasks.

Previously, modifying these host-level components meant altering task definitions and coordinating with development teams—an inefficient process at scale. The new model grants infrastructure engineers independent control. They define daemon task specifications with dedicated CPU and memory allocations, then attach them to capacity providers. ECS guarantees the daemon launches first on any new instance and drains last during updates, maintaining continuous data collection.

Architecturally, this hinges on a new `daemon_bridge` network mode, allowing the daemon to communicate with application containers while staying clear of their network stack. Daemons can be configured as privileged containers with host filesystem access, which is necessary for many monitoring and security tools. During a daemon update, ECS executes a rolling deployment, provisioning fresh instances with the new version, migrating tasks over, and then retiring the old instances. A configurable drain percentage controls the rollout pace.

There is no extra charge for the feature; you pay only for the compute the daemon tasks consume. It is available now across all AWS regions.

Source: AWS