Cloud provider limitations
The table outlines technical specifications and limits that are specific to the cloud provider.
| Cloud provider | Functions per CDF project | Concurrent calls | Schedules | CPU cores per function call | RAM per function call | Function call timeout | Function call data payload size |
|---|---|---|---|---|---|---|---|
| 100 | 100 per function | 1000 per project, 100 per function | default: 1.0, maximum: 3.0 | default: 1.5 GB, maximum: 5.0 GB | 9 minutes1 | 9 MB | |
| Azure | 100 | 100 per function | 1000 per project, 100 per function | 1.0 (not configurable) | 1.5 GB (not configurable) | 10 minutes | 36 kB |
| AWS | 100 | 100 per function | 1000 per project, 100 per function | 1.0 (not configurable) | 1.5 GB (not configurable) | 10 minutes | 240 kB |
1On 2024-02-01, the function call timeout for projects running on Google Cloud Platform (GCP) was changed from 15 minutes to 9 minutes.
Manage quotas
To manage quotas effectively, consider adopting the following guidelines:
Utilize schedules
Each Cognite Function can manage up to 100 schedules. To run the same function with different configurations, use the data field in the schedule. For example, if calculations vary across multiple sites, include the logic in the function code and trigger the correct version based on configuration. This combines multiple sites into one function, reducing the need for separate ones and optimizing quota use.
Regular audit and clean-up
With time, some Cognite Functions may become obsolete or unused, consuming quota unnecessarily. Audit your deployed functions regularly to ensure only active, relevant ones remain. Remove outdated functions to free up quota for new tasks and improve resource efficiency.
Governance
As the number of Cognite Functions increases, use version control systems (e.g., GitHub, Azure DevOps) to manage code and deployments. The CDF Toolkit simplifies deployment management. Assign owners to functions to ensure accountability during audits and clean-up.