amue wrote in Fri Apr 17, 2020 9:47 pm:Lag is not aircraft specific. Lag is network inherent.
With the current multiplayer implementation in FlightGear you have two choices:
Since especially for military flight operations (e.g. formation flight or dogfights)....
For me it's the FGUK Saturday night flights that truly can stress MP and show problems with variable latency because there can often be more than 10 players in close formation; OPRF is not often as much about formations; except for tanking and variable latency isn't something that I've often had a problem with probably because mostly ORPF events use a dedicated single MP server.
I've measured roundtrip time as upto 2000ms; typically ~1700ms on my system running two FG instances. This is with a 60ms ping to the FGMS - using the same server on different (incoming) ports.
The best option to mitigate network lag is firstly to all use the same server (as this removes server->server lag) and then to adjust the lag settings in the Multiplayer menu (see
http://wiki.flightgear.org/Real_Time_manual)
I did some brief testing and analysis of flights with pilots from UK, Europe and USA with two cases, firstly all using the same server (located in either USA or Europe) and secondly using the Geographically closest server. Typical USA->Europe transmit times are around 90ms, but did peak higher. All pilots using the same server works out at Pilot Ping[1] -> Server -> Remote Pilot Ping[3]. So for a USA based server this works out as
* USA -> Europe : [1] 30ms -> [3] 150ms
* USA -> USA: [1] 30ms -> [3] 45ms
* Europe -> Europe: 40ms -> 60ms
With multiple servers the local performance is better - but what I observed to be happening is that the times became very much more variable because Pilot Ping[1] -> Local Server -> remote server [2] -> Remote Pilot Ping[3] - overall the transmission time can be lower because [2] is often higher due to improved bandwidth but overall this appears to work out a lot more variable because there are more network connections and server processing.
This worked out as
* USA -> Europe : [1] 30ms -> [2] 95ms -> [3] 50ms
* USA -> USA: [1] 30ms -> [2] 18ms -> [3] 45ms
* Europe -> Europe: [1]40ms -> [2] 12ms -> [3] 60ms
So pretty much all comms will have a higher latency when pilots flying together are on different servers because these servers have to transmit data between themselves; so really the star topology is more about federating data rather than performance.
So generally my advice is for group flights to pick a single server and use it; then you can adjust your lag settings to that server and it can work out well without too much fluctuations - but obviously the whole MP comms could benefit from a total rewrite but that's unlikely anytime soon.