Just thought I would bump this with a status update. The state of multiplayer seems much improved in the upcoming 3.8. The ping calculation had some long-standing bugs that have been fixed. I haven't had connectivity issues at all recently, between my house, my work, and the 4 VPS standalones I have running now. In-game lag is another issue still unfortunately, as it is still reportedly worse than retail. But I think we have made some steps in the right direction, and getting recent confirmation that the lag is indeed still pretty bad would be helpful. For confirming this, I specifically set up a retail standalone using an unpatched release build, which I will attempt to leave running full time, and only restart to update to a new revision. I will try to keep it on what seems to be the fastest standalone (by pings from FS2NetD or my house), CentOS was winning but the openSUSE one I just set up is responding with faster ping times using a Debug build. So I may move it to that one. Just pick one with (Release) in the name. No need to worry about port forwarding to host, unless you want to run a mission I don't have. All of my standalones are currently retail, with the multi-mission-pack. There may be more added in the future.
jg18 has gotten pretty close to native integration for SquadWar. But he's had to take a break from that to focus more on RL related things. I'd like to give a huge thanks for the work he's put in so far on this and other multiplayer aspects though. It's been hard to get coders willing to put work in for multiplayer and he has done more in this specific area than many recently.
After many years of annoying me, I fixed FS2NetD to display raw ping values when you hover over the ping bars graphic on the website games list. This has revealed that the ping values are substantially higher than one would see in game. Given that my standalones are all over the country, but FS2NetD's pings are universally higher than I see in game, I think this is another ping calculation bug in the FS2NetD server and not just that it is farther away geographically from all the servers than I am (pretty sure that's not possible). The relative scale still matches up though, so it can still be relied upon as much as it could before you could see the numbers.
If someone wants to have a go on a test some evening, drop on IRC in #multi, #freespace, etc, or the Discord server #freespace and see if I or anyone else would like to go for a round or two. I'd love to put the standalone servers through some real testing.