Hey there! We've found the following might help you get unblocked faster:

• 📚 Railway vs. Vercel

• 🧵 Very high and inconsistent round trip times to backend from end user

If you find the answer from one of these, please let us know by solving the thread!

We’ve been waiting for more than 11 hours for a response to the issue we’re having.

Hey there Motorflash,

Sorry for the delayed response, been working with the team to get a resolution. We have a core fix coming that would address the latency instead of updating the peering table one by one.

Can you run this and share the results here please? https://docs.railway.com/reference/network-diagnostics

Hello, I’m attaching the diagnostic.

We need to know when the solution is going to be applied so we can stabilize the systems, as we have a frontend down because of this issue.

I am a bit confused, the front-end shouldn't be down because of this.

It looks like the diagnostic is showing that at least from your machine, there isn't too much latency. What region is your vercel deployment on?

Hello,

Paris, France (West) - eu-west-3 - cdg1

Hello!

We're acknowledging your issue and attaching a ticket to this thread.

We don't have an ETA for it, but, our engineering team will take a look and you will be updated as we update the ticket.

Please reply to this thread if you have any questions!

Noted, is it possible to put it in Amsterdam? I have also attached a ticket such that our Network Engineer can see this case as well.

We have this options available in Vercel, which one is better?

Paris, France (West) - eu-west-3 - cdg1

Stockholm, Sweden (North) - eu-north-1 - arn1

Dublin, Ireland (West) - eu-west-1 - dub1

London, United Kingdom (West) - eu-west-2 - lhr1

Frankfurt, Germany (West) - eu-central-1 - fra1

Frankfurt, Germany (West) - eu-central-1 - fra1

Also our network engineer got back to us, there is a global fiber cut that is impacting a few data transit providers. Our upstream vendor says that they expect a fix to be live by Monday. In the short term, you can consider deploying your front-end on Railway to minimize latency.

Hi, tonight from 23:45 to 00:45 CET there is a peak in memory usage and a cut of network egress, affecting the service. You mentioned that by Monday the fiber cut would be resolved, is this the cause of the behaviour? Can we assume it's already fixed?

Apart from connecting from Vercel to Railway (currently yet in eu-west-3 - cdg1) we also connect to mongodb atlas deployed in AWS Spain (eu-south-2).

Hello!
We're still waiting to hear back from our vendor according to our network engineer. We expect a response today.
Will update you here when we get info from them!

That being said, I have seen multiple clients experience much faster database services by hosting their mongo instance on Railway. My suggestion would be moving that mongo atlas db from AWS to your Railway project and connecting over internal networking!

🛠️ The ticket Global Connectivity Enhancement has been marked as todo.

Any news about the network issue?

Things should have been resolved and good to go! We also implemented a global backbone which should improve routing and latency.
Are you able to test and see if you're still having these issues?

Today we have a network latency between our instance in railway and the database in mongodb. The mongodb atlas is deployed in AWS Spain (eu-south-2). The issue was between 9:20 and 10:20 CET where no egress traffic was registered in your metrics. Could you please let us know if today you have any network issue?

Hello we have not had any network related issues as of late.
I've seen a few people report high/inconsistent latency using mongodb atlas hosted on other services. I'd suggest trying to move that database on your project and using internal networking! Things should be at the speed of localhost then

✅ The ticket Global Connectivity Enhancement has been marked as completed.