Hard to say without seeing the project, but to eat up 8s it must be doing some initialisation with e
Hard to say without seeing the project, but to eat up 8s it must be doing some initialisation with external API calls? Or is there some cold start or Turso's side related to accessing Turso from different regions (I haven't used Turso so I don't know how it operates).
From all my experience, I can reach workers on the far side of the world consistently around ~200ms, cold start or not.
So it could be related to doing some expensive (long running API request) initialisation on each worker request? If so, you could try firstly "caching" that initialisation in global variables so it is shared between requests to the same worker instance. And secondly caching initialisation responses locally in the datacentre using the Cache API so it can be retrieve very quickly even for a new worker instance.
However if your code is indeed just a "simple POC" with no such long running initialisation then I've got a feeling that the delay may be on Turso's API?
From all my experience, I can reach workers on the far side of the world consistently around ~200ms, cold start or not.
So it could be related to doing some expensive (long running API request) initialisation on each worker request? If so, you could try firstly "caching" that initialisation in global variables so it is shared between requests to the same worker instance. And secondly caching initialisation responses locally in the datacentre using the Cache API so it can be retrieve very quickly even for a new worker instance.
However if your code is indeed just a "simple POC" with no such long running initialisation then I've got a feeling that the delay may be on Turso's API?
