At 11:30 UTC we started getting tickets about customer's applications not responding.
We started investigating. It looks like the network or the reverse proxies are responsible for that.
EDIT 12:46 UTC: we are experiencing abnormal new connection rates on public reverse proxies.
EDIT 12:50 UTC: we found the responsible application for this new connection rate and are mitigating it.
EDIT 14:19 UTC: Load balancers have been upscaled so they can handle more traffic. Performance is back to normal since 13:12 UTC.