Important update notice
As part of our ongoing commitment to provide you with the best possible experience, we are enhancing API Mesh's infrastructure to run at the edge in 330 global locations.
Multiple partners are already using the updated version of API Mesh and have seen significant improvements in performance and reliability.
Due to compatibility limitations, certain features, such as replaceField transforms are not available in edge meshes.
Why the change?
- Improved response times and enhanced API performance
- Easier integration with industry-standard tools and third-party products
- Reduced potential of degraded performance caused by other hosted meshes
- Better security and compliance
- Increased observability
What does this mean for you?
- Continuity: Legacy meshes are disabled as of December 03, 2024. You must use the
edge-graph.adobe.ioURL to ensure your application does not face any downtime. - Action Required: Upgrade your aio-cli plugin to the latest version. You will see minor changes to the CLI experience to display a new "mesh" endpoint. This means your mesh is provisioned to the new API Mesh infrastructure. Going forward, you will need to use the "mesh" endpoint for all applications to ensure functionality past the completion date. See create a mesh for more information.
Edge meshes can take slightly longer to build. Wait two or three minutes before running the aio api-mesh: status command. Consider using local development for testing and development purposes.
Fastly and edge meshes
If you intend to use Fastly, follow Configure Fastly for edge meshes.
Key Dates
Update Start Date: August 15, 2024 Expected Completion Date: October 15, 2024
If you have any questions or concerns, contact our team directly, by emailing nrobichaud@adobe.com and pganapat@adobe.com or by joining the #app-builder-community Slack channel.
Performance testing
When performance testing edge meshes in API Mesh, you need to account for cold starts to get an accurate measurement of the performance.
If applicable, you should use the Connection: keep-alive header described in Optimizing edge mesh performance.
Alternatively, you can manually warm the cache using the process described in Performance testing.
