Hi everyone - I opened a case for this on Central Station maybe 10 days ago and have had a bit of input from the community but nothing from Railway itself. Original thread: https://station.railway.com/questions/my-sql-latency-issues-cdeea77b

Im seeing odd, variable latency from MySQL in multiple railway projects located in the Europe zone. My expected latency over a local interface is 1-1.5ms and my applications running are consistently seeing 5-7ms round trips.

A ping test confirms of that, 5-6ms is round trip latency which is very high for assets in the same physical facility.

By creating a new project I could see a reduction in this latency and it doesn’t appear consistent across different services in the same project. Modifying one of my services and causing it to redeploy brought average latency down but with huge spikes in ping during testing. The same application running in production conditions (small app) saw sub-ms response times with MySQL as another host on a single physical Docker server.

I’ve replicated this across several images, and shown that the issue is not consistent with the application nor the underlying build - I don’t believe the issue is specific to the docker image (I tested a stock bookstack image from linuxserver and it showed the issue too).

I believe this may be an infrastructure quirk or issue but I don’t have access to debug it further and I’m hoping for some help from Railway staff to identify and fix the issue. For chatty applications like Laravel 5-7ms per call plus additional latency on the first connection makes a meaningful difference to the performance of my apps.

I would also welcome any input on how to flag a ticket into the private queue with railway - I understand the logic of making non sensitive stuff public to help others but in my case I included semi private screenshots of my app structure and making it public for me without my asking also meant that I didn’t get a response from Railway at all.