Hi! (apparently tilda crosses out, so here it is again, apologies for the triple post)
We use PlayFab MPS for hosting servers for a 1v1 / 2v2 fighting game, as well as matchmaking / general backend etc. Broadly MPS has been very effective for us so far and straightforward to use!
However we had a public beta late last month with approx. 10k participants and had an issue crop up with certain players where their pings to certain playfab regions were much higher than expected. IE. Players located in the Los Angeles area who had pings of 150+ to West Us and West Us 2 beacons (and to the servers as well if they forced matchmaking to only place them there). The issue was persistent and lasted the whole beta weekend, and in certain players cases functionally prevented them from playing online as they had extremely high pings to all available server regions, despite being located near them.
After some investigation, if affected players used something like https://visualtraceroute.net/ to see the route their game packets were taking we noticed that their packets were being routed very oddly (ie LA to West Us, going from LA to the midwest, then to San Jose) and then back to the server / beacon.
For the player in the above image, they have Google Fiber in LA, and were able to use a VPN they have through their work to play at expected pings (<20-30 to West Us), so definitely seems like something with their ISP just routing packets to PlayFab very poorly. However it wasn't specific to ie just Google Fiber in LA, and affected people in Phoenix / Seattle / West Canada and some people in the upper midwest / Minnesota area, etc.
If that's the case, is there anything we can do to help resolve this issue for players? It wasn't widespread but it more or less prevented the affected players (approx 50) from playing in ideal conditions and didn't resolve throughout the beta weekend. It seemed to be primarily affecting the West and Central US server regions, as we had no reports from players close to East servers encountering the issue.
Additionally, we've been noticing a similar issue for players who have 'normal' pings to servers, where when connected to a server their ping to that server suddenly inflates to about double of their normal value (ie 60 to 120), (40 to 80-90 etc), and does perform the way you would expect with that inflated ping. It can remain that way for around 15-30 minutes and doesn't affect everyone connected to that server. It's weird because while experiencing the inflated udp latency you can ICMP ping / tracert the server ip and get normal values, but if you udp ping the QoS beacon you get the same value the game is reporting.
I'm unsure if this is related to the regional routing issue but feels at least of a kind to me, and would be curious if anyone knows what may be causing / what could be done about it?