…and before anyone jumps in with more talking points from CI¬G — a company that has demonstrated no working knowledge of multiplay gaming in general and networking in particular — no, “server meshing” will not solve that problem. In fact, it could just as likely to make it worse if it was attempted.
The problem is fundamentally one of keeping the instance and everyone in it updated with the information needed to see the world around them. The stupendous but wholly ignorant idea that has been floated on top of CI¬G PR is that, through meshing servers, you get around the processing limits implied by such a division by having different servers deal with subsets of the workload. Double the number of players = four times the communication needed (because it's an O(n²) problem to keep everyone updated) = mesh four servers to deal with it… right? Except no, because this assumes that player 1 will only deal with instance-partition 1, and ignores that the whole point is that they also see players 35, 88, and 159 in the other three partitions. So all those updates still need to happen, and all partitions need to be aware and updated on the state of all players. Perhaps not the entire state, so some gain could be had by filtering out and not sending that data all over the place, but for the exact same reason, that data would not be a huge part of the workload to begin with. And that would count against the increased overhead of keeping the meshing going and figuring out what other partitions need and need not be updated with.
They're trying to parallelise an asymmetric and stateful problem that therefore cannot be parallelised, and they're seemingly ignoring the overhead costs even if that could be done.
So server meshing would “solve” the problem of not being able to deal with 2× the players (i.e. 4× the work) in a single instance by spinning up three more servers and making them deal with 15× the workload. Why 15 rather than just 4? Because, again, the work can't be cut cleanly; each of the four servers will still need to know most of the things about those other players — not all, but most, and especially the parts that already create the majority of the workload — and a whole bunch of syncing overhead on top of that. Congratulations, instead of having 1 server be 300% over capacity, your server meshing has made 4 servers be 275% over capacity. That's a whole lot of overworked servers for no gain (everyone is still stuck on an overworked and stalled server) when the intelligent and sensible solution — and therefore the one that CI¬G is fundamentally, almost genetically, incapable of producing — is to decide on a hard limit to the partition size and… you know… design a game that operates within those limits.
But that requires forethought; it requires planning; it requires letting limits drive creativity; it requires saying “no”. It will never ever happen with this company and these people in charge.