I know a lot of financial decisions have to be made when estimating server load at a game's launch, but is there really no other way? Nearly every big name MMO, or always-online game, is an absolute mess on release. How many potential players/subscribers are they losing by not having the game be accessible while it's at its most momentous?

It seems like it would make more sense to take the financial hit in the beginning and load up on servers. At least then you can adjust down if the population decreases rather than trying to catch up, and rendering the game unplayable.

How crazy is it going to be if the same thing happens with Elder Scrolls Online? We'll get that inevitable and familiar statement, "We are thrilled with the response to Elder Scrolls Online. As a result, however, server load is above our expectations blah blah blah." while thousands of people are soured on the experience.

Here's hoping FFXIV is the last time this happens. You gotta' know what to expect by now, big name MMO makers.

P.S. I am completely ignorant on the topic, and have no idea what I'm talking about.

It's going to be the case with elder scrolls online. You always expect that the next big launch will be better, but it's not. Practically EVERY MMO launch is bad; just some are worse than others. Comparative to other big MMOs released over the past 15 or so years we've had MMOs, FFXIV's was decently okay. It's still POSSIBLE to play the game, when you actually get into the game the game isn't broken.

If we're still having these issues a week from now I'd call it a bad launch, but still not the worse. Don't forget the ridiculously bad rep that FFXIV has had ever since it originally came out. Were YOU expecting it to actually be that big of a hit? I wasn't, and neither were the developers.

Splitting up the servers between JP and NA/EU didn't help; obviously there's going to be more space on the JP servers, hence why they're not really having a terrible time of it.

I doubt anyone involved could have guessed it would have been the NA servers that would be packed to the brim. The general attitude towards these games has been pretty dismissive in the west as of late, so I can't put all the blame on them for being caught off guard. To their credit, they quickly came out to say they were adding more servers this week, instead of just riding it out until things settled down, or leaving people on the hook while they tried to figure it out in the background.

Despite Brad's middling view of the game, he brought up a good point in the latest Bombcast. If Square was to overestimate how many players they needed to support, then they would be spending huge amounts of money on bandwidth and hardware that wasn't being used. I don't think anyone wants it to be true, but sacrificing a few early players is preferential to overspending on network infrastructure.

@zeik: In fact, Yoshi-P practically said as much at PAX. He said after v1.0's bad release, everyone but the diehards would stay on the fence until it was deemed safe. But the generally positive word of mouth coming out of beta3, and probably more than a little bit of people just wanting a new MMO, turned everything on its head.

This happens with just about every MMO. At this point it's easier to just wait a week or two for the servers to be stable.

I'm just going to say its not an easy problem to fix. Every server they put up costs them exponentially. If you buy a ton just to have for release your essentially wasting money.

And let's not forget that after the first month ends, they'll probably lose half their server load

Yeah, this has to my knowledge happened with every single MMO launch, and it'll probably keep happening until someone figures out a secure way to do some kind of bittorrent-ish thing were players are also acting as servers while playing the game or something.

P2P servers have their own set of problems. I don't think that's the solution, unless it was only a temporary thing to alleviate launch congestion.

Once once you lose that initial burst of customers, you open up a new problem.

All those servers now have low populations and people are unable to find people to play with. Even if you have a -very- healthy population, if stretched thin, every server's going to look like a ghost town. How do you fix that?

Merging servers. However, this causes issues too. How is Sephiroth on server 1 going to share names with Sephiroth on server 2? Which Sephiroth is going to have to change their name to Sephiiroth?

On top of the naming nightmare, critics will eat up the "We are merging servers 1,2,3,4,5,6,7,8 to server 9,10,11" messages. Even if servers 9, 10, and 11 will house 9 million people, the naysayers will start saying, "The game is failing, look at those mergers!"

The best decision is to have these crappy launches. Eat the first few weeks of subs and say, "Hey, that time was free. Thanks for hanging in there." You trickle players in as other players leave and you pique people's curiosity by having reports of a game so popular, not everyone can play it.

Once once you lose that initial burst of customers, you open up a new problem.

All those servers now have low populations and people are unable to find people to play with. Even if you have a -very- healthy population, if stretched thin, every server's going to look like a ghost town. How do you fix that?

Merging servers. However, this causes issues too. How is Sephiroth on server 1 going to share names with Sephiroth on server 2? Which Sephiroth is going to have to change their name to Sephiiroth?

On top of the naming nightmare, critics will eat up the "We are merging servers 1,2,3,4,5,6,7,8 to server 9,10,11" messages. Even if servers 9, 10, and 11 will house 9 million people, the naysayers will start saying, "The game is failing, look at those mergers!"

The best decision is to have these crappy launches. Eat the first few weeks of subs and say, "Hey, that time was free. Thanks for hanging in there." You trickle players in as other players leave and you pique people's curiosity by having reports of a game so popular, not everyone can play it.

Exactly, i work in IT and have an infrastucture team who deal with this kind of thing on a daily basis, everyone keeps harping on how they should have been more prepared and how it happend last time so they should have learned, but its just not that easy. You can spend millions building up servers that are only needed in reality for a few weeks and then become infrequently used, its a no win situation and as a customer yeah, im pissed when i cant get in but i understand that its not as simple as "Square have no idea about mmos".

Is the game setup in a way that several instances can be spawned of a zone? If that is the case I don't see how they could have such a rough launch. The instances could all run on different (virtual or physical) servers, and you can make the game server (I think they call them shards to mark the difference?) as big or small as you want purely from a design standpoint, there are no real technical limitations anymore.

Just set up your game at one of those cloud providers like microsoft or amazon and scale the capacity based on demand. Yeah, you are still paying for the capacity but because of the unexpected success you are also receiving a ton of money, and you don't have any abandoned servers after launch or empty shards.

IIRC this is how a ton of mmo's like Tera and Neverwinter Online did it, and those seemed to launch relatively smooth.

Am I missing something here?

MMO launches are complex where it isn't that any particular vendor didn't prepare for it but that parts of the Internet and infrastructure can't prepare for it. Companies that run parts of the Internet have "peering agreements" which could be as "simple" as they do not charge each other money as long as their bandwidth to each other is relatively equal. When something like an MMO release happens, some company out there is seeing a huge spike in bandwidth in both directions where they will end up being charge a ton of money unless they either find more traffic or start throttling things.

It isn't just the "game servers" but the entire infrastructure. Web servers as simple as the media and information and even phones are being hammered. Unless there is a radical change in the architecture this will happen again and again for popular game releases.

@jozzy: Just that it is terribly complicated to do so :P. Not just setting up but also maintaining and updating the servers.

I was an alpha/beta tester on Neverwinter and the launch was far from smooth, the game's severs kept crashing and thanks to it sharing the same cluster of virtual servers as Star Trek online both games would go down at the same time anytime the log in count got too high. When it did work there was at least a 3 day period when anyone outside of the US couldnt get logged in and then there was the crazy lag spikes...

Dont get me wrong, i love Neverwinter and Star Trek Online and as mentioned, i understand the issues which is why i didnt turn into one of those flamers on the forums and facebook page threatening to sue/kill members of Cryptic. Ironically the stress test and open beta was okay, it was the official launch that killed the service and much like this game they spun up additional servers as needed but had downtime of their own.

It took them roughly a week to fix it, and while logging in week one was a bitch... you could still play the game. Since yesterdays server update and maintenance I have had no log in issues at all personally. Also they gave everyone who bought the game and had an active account in week one a free 1 week extension of their sub. I have seen plenty of launches go far far worse than this one. Anyone who quits over week one log in issues in an mmo is either someone who was going to quit anyway, or this is their first mmo and they just had no idea what they were getting into.

Agreed, mmo launches will never be perfect regardless of how prepared the developer thinks they are, you just cant guarantee stability until you have actually had the full number of users log in and overload the server a bunch of times. Stress tests as good as they are still dont simulate real world numbers since they are controlled to a certain degree, i think a weeks turnaround is pretty good but it also helps that i love and have loved the game since the AAR beta. I never felt that way once about 1.0