Cloud Composer has the following API quota limits.
|Write requests per project||1,500/minute, 25,000/day|
|Read requests per project||1,000/minute|
Most API calls cost 1 unit to perform, but some are more expensive. The following APIs
cost 100 units to perform:
The following tools are built on top of the Cloud Composer API, and use your Cloud Composer API quota:
Cloud Composer utilizes other Google Cloud products. These products have project-level quotas, which include quotas that apply to Cloud Composer use. Some services are required to use Cloud Composer, such as Google Compute Engine and Google Cloud Storage. Other services can optionally be used with Cloud Composer when setting connectors in Airflow.
The following services, which enforce quota limits, are automatically (required to be) used to create Cloud Composer environments.
- Deployment Manager
- Kubernetes Engine (which relies on Compute Engine)
- Cloud Storage
- Cloud Pub/Sub
- Cloud Logging
- Cloud Monitoring
- Cloud Build (if using PyPi dependencies)
- Google Container Registry
- Google Artifact Registry
Deployment Manager manages other Google Cloud resources needed by Cloud Composer. Each
delete call translates to one Deployment Manager write query and multiple read queries where
the Deployment Manager quota applies.
Airflow scheduler and workers run inside a Kubernetes Engine cluster. Each Cloud Composer environment creates one cluster. An environment by default has three Compute Engine VMs, though this number is configurable during or after creation. The cluster effects the Kubernetes Engine quota while the number of nodes effects the Compute Engine quota.
Cloud Pub/Sub is used for communication between Composer Service and the Airflow deployment we create. Each
gcloud composer environments
update call creates two Cloud Pub/Sub topics and two Cloud Pub/Sub subscriptions. The Cloud Pub/Sub quota applies.
Cloud Logging and Monitoring
Cloud Logging is used for Cloud Composer logs and streaming Airflow logs. The Logging quota applies.
Cloud Monitoring enables you to collect and visualize your Airflow metrics in a custom dashboard. The Monitoring quota applies.
Cloud Build allows you to build custom docker images when installing PyPi packages. The Cloud Build quotas applies.
You can use Airflow operators with Google Cloud services. Any service used in a Cloud Composer environment is subject to that service's quota.