8 days ago
Since somewhere between 8-11AM PT on 9/16, our network performance between Vercel (CA, SF region) and Railway (CA region) has got about 7x poorer. Before we were getting a P50 of ~20ms, now ~140ms. Our private network services within railway continue to have steady and good performance, so we believe this to be the networking.
We have no deploys or configuration changes at that time that could have caused the problem. Did anything change on your side?
4 Replies
Status changed to Awaiting Railway Response brody • 8 days ago
8 days ago
For low-latency connectivity between your app and database, we recommend running both in the same Railway project and environment and connecting over Private Networking.
Connections from Vercel reach Railway over the public network. Having both services in California doesn’t guarantee a direct or low-latency route between them, and we can’t support latency troubleshooting for this configuration over the public network.
If you move the app alongside your database and still see elevated latency over Private Networking, send us the updated measurements and we can take a look.
Status changed to Awaiting User Response Railway • 8 days ago
8 days ago
I'm sure this is an extremely common configuration for your customers, so much so that you maintain an integration with Vercel to help make it easier. It's not reasonable to throw your hands up in the air and claim that you provide no expectation of decent public network performance. And regardless, our users would experience our services through your public network even if we moved everything into Railway.
Status changed to Awaiting Railway Response Railway • 8 days ago
8 days ago
We’re looking to deprecate the Vercel integration entirely. The integration’s availability doesn’t mean we can guarantee latency or support troubleshooting the network path between Vercel and Railway.
Your users would still access the application over the public network, yes. The distinction is that your application’s database connections would stay on Railway’s private network. That keeps the public network out of the app-to-database path, where repeated round trips can compound latency.
For low-latency app-to-database connectivity, our recommendation remains to run both in the same Railway project and environment and use Private Networking.
Status changed to Awaiting User Response Railway • 8 days ago
a day ago
This thread has been marked as solved automatically due to a lack of recent activity. Please re-open this thread or create a new one if you require further assistance. Thank you!
Status changed to Solved Railway • 1 day ago
an hour ago
Randomly, this seems to have fixed itself at 11:30pm PT yesterday. Our latency is now behaving as expected. Sharing in case useful.
Status changed to Awaiting Railway Response Railway • about 1 hour ago
Status changed to Solved Railway • about 1 hour ago