Cloud Run Limits
Cloud Run is in Alpha, which means APIs may change in any version and documentation is not yet finished. See the changelog to stay up to date with breaking changes.
The standard GCP Cloud Run quotas apply (see here), most notably:
- Concurrency: By default, Remotion Cloud Run is deployed with 0 concurrency. This means that each request will get its own instance. This can be changed in the GCP Console, however concurrent requests on one instance has not been tested by the Remotion team.
- Instances: The number of Cloud Run Service instances that can be created in response to requests. Configurable, limited to 100 at most. It is possible to request a higher quota limit through the GCP console.
- Memory: Configurable, limited to 32GB at most
- Execution limit: Configurable, at most 60 minutes
Steps to increase Instance quota
Navigate to Quotas within IAM and select your Remotion project.
You are able to make a request for an increase in Instance limit per GCP region. Select each region required, ensuring the value in the Quota column is Instance limit per region. In the top right corner, select Edit Quotas.
Follow the prompts, using these points below as a guide. There is no guarantee that the example descriptions will work future requests.
- The 'Instance limit per region' quota is part of the Cloud Run Admin API.
- Example request description - "Looking to ensure no wait times for our users. The Cloud Run service we are running cannot have any concurrency, and therefore we rely on spinning up extra instances for multiple requests."
- Example intended use case - "Rendering videos for users, using Remotion (https://remotion.dev). Each request to a service would render a video and store it in Cloud Storage, before finishing."
- Example intended usage pattern - "I expect bursts during business hours in Australia."
- When asked if the container supports concurrent requests, input a 0.
- Example of how I arrived at the number of instances being requested - "Looking to double the existing quota."