At the ESA Living Planet Symposium last week, we were happy to welcome many new users and people who were interested in the platform. While this is of course very good news, we also noticed that unfortunately the increased usage also resulted in the Terrascope backend being unable to keep up with demand. As a user, you probably saw some very slow requests as a result, or errors.
Therefore, we are currently rolling out some changes to improve upon this, and allow more users on the platform. If you experience issues because of this, or if your project requires a specific capacity or SLA (service level agreement), please do let us know! We can always consider how we can accomodate specific needs.
These are the measures that are currently being implemented. Some of them will apply to the full platform, and some for the Terrascope backend:
- The Terrascope backend capacity will be increased by setting up a second instance. This will result in a highly available ‘active-active’ setup. This measure both increases the number of concurrent users we can handle, and the overall availability.
- Caching for certain requests will increase overall responsiveness of the platform. Metadata requests for collections and processes will be more consistent.
- Rate limiting will limit the number of requests that can be made in a given time frame. Sometimes users write code that floods the backends, this will now result in an error saying '429 Too many requests".
- The number of concurrent synchronous processing requests will be limited to 2 per user. This may be made dependent on the type of subscription in the future.
These changes will still take some time to roll out. Do let us know if you experience issues in the meantime or have questions!