So what was the point of the server upgrades when I can't continue the MSQ?

[Saccharin]

At least for EU login server did crash and long ques for popular servers. No idea about GW2 expansion but the hype for that game really died so is no wonder.

Except for those few mega popular servers the rest went fine.

The way blizzard solved this was to essentially borrow a technique from mega server, thee realms are basically mega servers now.

When you move into a zone and it's too busy a new instance is created therefore reducing the burden on the servers. This worked flawlessly for people on Legion release.

The problem with FF14 is that it uses very poor server organisation.

Edit, turns out they use instances now!

SE really need to get their act together and write proper server code. Sadly the Achilles Heal of this game (2.x+) has always been it's poor, limited server code.

[Claviusnex]

Except for those few mega popular servers the rest went fine.

The way blizzard solved this was to essentially borrow a technique from mega server, thee realms are basically mega servers now.

When you move into a zone and it's too busy a new instance is created therefore reducing the burden on the servers. This worked flawlessly for people on Legion release.

The problem with FF14 is that it uses very poor server organisation. They, as far as I know, keep everyone on the same instance so you can only get in so many, which is odd considering each zone is instanced, and not seamless like WoW.

SE really need to get their act together and write proper server code. Sadly the Achilles Heal of this game (2.x+) has always been it's poor, limited server code.

I agree with you that whoever SE has doing the server and system parts don't seem to have a good handle on their environment. Of course it could also be a management problem as this type of support isn't "Content" nor considered sexy so often gets the short shrift when it comes to funding.

As to design it shouldn't have to be that complicated. All they need to do is setup instances to run in their own server pool. As instances are generally going to be well known in their system resource requirements it should be straight forward to determine how many instances they can run on a system image. If the pool gets full then add more system images to the pool (could be hardware or software). When things drop off remove the extra images. It is not rocket science as this type of thing is done everyday for eons on any large, high transaction system.