Similar to this now hidden thread: https://discord.com/channels/713503345364697088/1359568597491388606

We migrated almost al our services to Metal, but are experiencing very bad response times, it seems worse on services connected to other database services.

Attached an example of the identical service(s) where I ran a simple quantity mutation on a cart 3 times. On GCP they are all ~500ms on metal 2-4s, so 4-8x slower! This project uses a private database URL. Project ID: 0b36733d-8a0b-49ac-81c7-a9ba22c3b15a

We also have another project where a simple json fetch from database records, though a big payload, returns on GCP in 0,9-1s and on Metal it takes 5,4-21s, 5-10x slower! This project uses a public database URL. Project ID: c6f6b68a-8b76-42d7-bd0a-65e794ac8fef

Interestingly enough, if I only put the database in the last project in the Metal region, there seems to be no issue. When I put the application service back to Metal region, the response times are worse again.

I am a bit surprised by these results, also since volumes on GCP should be '10x faster', not sure on which metric that applies? Just querying directly to the database service did not give any performance benefits on Metal.
Are there any major issues still with Metal I am not aware off? Will revert the services back to GCP for now.