The fact that it look over an hour is a bigger worry and implies they set the limit on the max number of servers to launch too low.( it's normal to have a limit, stops a DDOS attack from causing 1000s of servers to run ).
I'm just guessing here, but I wonder if there's a problem with certain types of load going up faster than linearly with the number of players, when those players are all in the same "place".
Wild guess: when you try to enter a location, it looks at all possible instances in that location, and then selects the "best" based on the various instancing factors, or a new one if all the existing ones are below some threshold.
So every new person entering a busy location first has to check all the existing instances, and then may sometimes create a new instance. As more people enter, the number of instances rises, so each new person takes even longer to add (and if lots of people are arriving at once, the instancing calculations may be invalidated by the time they're complete).
That means that the complexity of the instancing problem may be proportional to the *square* of the number of players, and that can lead to the load increasing very rapidly as the number of players in the same place increases. (So a few thousand people in one place will crash it, while a few tens of thousands spread out it'd barely notice)
Lead-up: servers are busy, but people are only joining the instances one or two at a time as they arrive. The load is high, and it takes a little while for them to sync in to the instances, but the server team are monitoring and spinning up servers as needed to cope.
Mass-jump: there's now a few thousand people all jumping to the same target system supercruise simultaneously. The instancing system has to cope with a load possibly hundreds of times higher than it was dealing with during the lead-up. Unsurprisingly, it falls over.
Recovery: those few thousand people all try to log in again. As they crashed out in hyperspace, they get recovered by the game to the "default" position in the new system. But there's a few thousand people now trying to log in at the same time to the same location, so this is basically the same instancing problem as the supercruise one that just killed the servers.
It may not even be a problem you can "just add more servers" to fix at this point - any individual player's instancing calculation may now be too complex and involve so much conversation between servers to work out that at some point adding extra servers might start making it worse by spreading out the data too much.
If I'm right about this, the solution is for people on mass jumps not to all jump to the same destination - pick a bunch of different systems in roughly the same direction.
