Thursday, June 11, 2015
Launch issues
This week The Elder Scrolls Online launched on consoles, and the launch didn't go well. Destructoid joked "The Elder Scrolls Online suffers totally unexpected launch issues", basically classifying the event as non-news: Pretty much every MMORPG has launch issues. The underlying problem is pretty basic: On normal days the peak concurrent users of a game constitute only a small percentage of the overall players, industry rule of thumb is about 10% for a subscription game. On launch day *everybody* wants to play, so the servers are overloaded. Basically launch day is pretty much indistinguishable from a distributed denial of service attack on the game. So launch issues are "normal". But is that really an excuse?
If you told a big internet company today that there was a distributed denial of service attack expected on their servers next month, they would be able to prepare and either negate that attack or at least mitigate the damage. So if everybody knows that launch day is a problem, why shouldn't it be possible to handle that problem as well? You would probably need additional server resources, but as you can rent cloud-based solutions that isn't impossible. It has a cost, but that should be weighed against the positive marketing effect of the really surprising headline of "MMORPG launches without problems".
At the very least a game company should provide enough login server resources to sort the mess into an orderly queue, because that by itself already solves part of the problem. If a player tries to login and gets an undecipherable error message, he will try again and again. If he logs in and gets told that he is in a login queue of 2 hours, he knows what is going on an might decide to give up and try the next day instead.
In the case of The Elder Scrolls Online it has also to be remarked that this is the second launch of the same game. Oscar Wilde would say that "To lose one launch may be regarded as a misfortune; to lose two looks like carelessness.". Online games aren't exotic any more. Even games that previously would have been considered single-player games in this day an age are often designed as online games, and thus frequently also suffer from launch issues. It is time that the industry finds technical solutions to this problem.
If you told a big internet company today that there was a distributed denial of service attack expected on their servers next month, they would be able to prepare and either negate that attack or at least mitigate the damage. So if everybody knows that launch day is a problem, why shouldn't it be possible to handle that problem as well? You would probably need additional server resources, but as you can rent cloud-based solutions that isn't impossible. It has a cost, but that should be weighed against the positive marketing effect of the really surprising headline of "MMORPG launches without problems".
At the very least a game company should provide enough login server resources to sort the mess into an orderly queue, because that by itself already solves part of the problem. If a player tries to login and gets an undecipherable error message, he will try again and again. If he logs in and gets told that he is in a login queue of 2 hours, he knows what is going on an might decide to give up and try the next day instead.
In the case of The Elder Scrolls Online it has also to be remarked that this is the second launch of the same game. Oscar Wilde would say that "To lose one launch may be regarded as a misfortune; to lose two looks like carelessness.". Online games aren't exotic any more. Even games that previously would have been considered single-player games in this day an age are often designed as online games, and thus frequently also suffer from launch issues. It is time that the industry finds technical solutions to this problem.
Comments:
<< Home
Newer› ‹Older
Companies do not invest into launch-day resources because it would a lot of money for something that everyone will forget few weeks later.
I still remember the horrible Diablo III launch-day fiasco. It was a HUGE mess and nobody was expecting that from Blizzard. Still... It happened.
Diablo III didn't suffer from that and it sold for inredibly high numbers, no matter if it was still version 1.0 wit a freaking ton of features still missing (which were added -in part- more than one year later).
Gamers have short memories and love being milked. They brag on forums, raise their pitchforks, claim they will quit forever... Then they come back.
I still remember the horrible Diablo III launch-day fiasco. It was a HUGE mess and nobody was expecting that from Blizzard. Still... It happened.
Diablo III didn't suffer from that and it sold for inredibly high numbers, no matter if it was still version 1.0 wit a freaking ton of features still missing (which were added -in part- more than one year later).
Gamers have short memories and love being milked. They brag on forums, raise their pitchforks, claim they will quit forever... Then they come back.
If every launch is a desaster - maybe "MMORPG launches without problems" would be read by the public as "not many people are playing"?
It's the same with all those "new product sold out in record time" headlines.
Maybe that's just what you do in marketing hype your product and present it as in huge demand
It's the same with all those "new product sold out in record time" headlines.
Maybe that's just what you do in marketing hype your product and present it as in huge demand
It's not exactly like a DDOS attack. In a DDOS attack, the goal is to ignore the offending connection as quickly as possible, where as in a login queue, all connecting sockets are customers you need to keep track of. And it's pretty easy to run out of sockets, they're a finite resource.
That said, though, the login part of the equation is decidedly far simpler than the game side in terms of server resources in general, so there is no real reason to NOT re purpose your extra resources temporarily to that. Even if you had to use game servers that were being held in reserve.
Having the login server resources to track and inform users of their queue status is job one. More important than to actually have somewhere for them to go once they DO log in.
That said, though, the login part of the equation is decidedly far simpler than the game side in terms of server resources in general, so there is no real reason to NOT re purpose your extra resources temporarily to that. Even if you had to use game servers that were being held in reserve.
Having the login server resources to track and inform users of their queue status is job one. More important than to actually have somewhere for them to go once they DO log in.
We also have a bad luck bias : we only remember the failure. GW2 launch was perfect for exemple, everybody was able to join the game without problem.
The problem with most MMO, is that player join ONE server. As long as the server is not fixe, Game dev can duplicate server and absorb the launch DDOS.
The problem with most MMO, is that player join ONE server. As long as the server is not fixe, Game dev can duplicate server and absorb the launch DDOS.
The GW2 launch was not perfect. They solved the issue of too many logins not by having more servers, but by only selling a limited amount of copies. This, in turn, meant that when, a week after the launch, I decided to play the game, it was unavailable for purchase, and remained so for weeks.
The end result is that I never bought GW2.
The end result is that I never bought GW2.
I was surprised that ESO had launch issues because they seemed to go out of their way to make sure that didn't happen back when the PC version released. During the PC launch, they had an overload server that allowed people to play even when the regular server was full. Of course they also spread everyone out a little more between the 5-day early access group and the 3-day early access group. Maybe they can't do the same sort of overloading with Xbox and PS4 though.
Post a Comment
<< Home