Re:Journey to Optimize Cloudflare D1 Database Queries
by arkh on Tuesday April 08, 2025 @02:21AM
(#43618857)
> This means a worker running in the same data center hosting your D1 databases might get ~10ms requests
This is already waaaaaay too much. Same datacenter I expect nano seconds round trips. Datacenter to a datacenter on the other side of the world? Maximum 300ms. It was what you expected of game servers two decades ago. I don't see what could eat more than a couple dozen ms on the server with current hardware to make a 10ms round-trip become "> 200ms" and consider this a service you'd pay for.
Re:Journey to Optimize Cloudflare D1 Database Queries
by manter on Thursday April 10, 2025 @01:38PM
(#43646272)
I think D1 actually requires multiple round trips, so the higher figures start to make some kind of sense.
I read a blog post about this but can't find it now. I did find this [release note](https://developers.cloudflare.com/d1/platform/release-notes/...) though.
They describe removing _at least_ two round trips back to the DC hosting the D1 instance.
I'm living in reality, where a globally available application needs to serve people all over the globe from Cloudflare's >300 data centers. This means a worker running in the same data center hosting your D1 databases might get ~10ms requests, but workers running on all other >300 data centers have to pay for the round trip to the primary data center whenever there is a cache miss.
The question you need to ask yourself is how come you failed to notice that. Do you have a good grasp on what you are doing?