Amazon MWAA Environment
Does ZopNight manage Amazon MWAA Environment?
Amazon MWAA bills each Airflow environment per hour by size class (small, medium, large), plus additional worker and scheduler instance-hours and per-GB metadata storage. Environments never stop on their own. ZopNight discovers them on the 6-hour cycle, attributes per-environment cost from Cost Explorer or CUR 2.0, and recommends class rightsizing.
Rules that fire on Amazon MWAA Environment
No active rule family targets Amazon MWAA Environment today. Rules that used to are retired, and retired rules publish no pages and fire no findings. Scheduling and permissions coverage are unaffected.
At a glance
| Field | Value |
|---|---|
| Scheduling notes | discovery, cost tracking, and recommendations only. |
Amazon Managed Workflows for Apache Airflow (MWAA) runs managed Airflow environments billed per environment-hour by size class, plus worker and scheduler capacity. Environments bill continuously, so oversized classes and idle dev environments carry fixed cost.
The environment-hour and its riders
MWAA’s base meter is the environment-hour, priced by size class (small, medium, or large), covering the scheduler, web server, and a baseline worker. Riders accumulate on top: additional workers bill per worker-hour as they scale with task load, extra schedulers bill per hour, and the metadata database bills per GB-month. The class choice is sticky and consequential: it sets the hourly floor and the capacity of every component, and it is chosen once, early, usually before anyone knows what the DAG load will be.
Orchestrators judged by their orchestration
ZopNight discovers MWAA environments automatically on the 6-hour cycle, attributes per-environment cost from Cost Explorer or CUR 2.0, and recommends environment-class rightsizing. The comparison that matters is class against workload: an environment running a handful of nightly DAGs on a large class is paying for scheduler and worker capacity sized for hundreds. Dev and staging environments get special attention, since an orchestrator that triggers jobs for eight hours a day still bills its environment-hours for all 24. The environment is always on even when everything it schedules is asleep.
Where Airflow environments overspend
Per-team environment sprawl leads: MWAA environments are easy to request and each carries the full hourly floor, so organizations end up with five environments doing one environment’s work. Class inflation follows a single bad week: bumped to large during a backfill, never bumped back. And minimum worker counts raised for a spike persist as permanent worker-hours after the spike resolves.
Environment inventory in the console
The MWAA console lists environments with class, worker range, and status. Airflow’s own UI per environment shows DAG counts and task concurrency, the workload side of the rightsizing comparison, and CloudWatch’s MWAA metrics chart worker scaling against the configured minimums that set the floor.