Following the shutdown of several servers for Flying Lab Software’s Pirates of the Burning Sea, the studio’s CEO has told Next-Gen exactly what went wrong.
Earlier this week, Flying Lab shut down seven of 11 of Pirates of the Burning Sea’s servers in an effort to condense the game’s population. The game launched in January.
CEO Russell Williams’ explanation of the server shutdowns could provide other MMO makers with the foresight to avoid a similar situation. Here’s Williams’ e-mail to Next-Gen in full:
The server merge is a result of a combination of things. When we were looking at the number of servers to roll out, we had a lot of different criteria to look at. Some worked out just like we’d expected, and some we were just wrong. Here are the pieces that went wrong:
At the beginning, we were very cautious about what our servers could support. Too cautious, in fact. Performance issues that we’ve seen have always been code, by which I mean that the servers weren’t being stressed, but the code was hitting problems handling too many users and would choke as a result. As it turns out, our servers can handle significantly more people than what we’d been assuming, in part because people in the beta had very different play styles than people on the live servers.
When I refer to play style, I’m talking about the way our players move in the game and where they congregate, which can create choke points. As I noted above, it turned out they spent more time in the choke points in the beta than in the live release. Of course we knew there’d be differences in the pre-release environment, but you go with the data that you have.
Pirates’ gameplay is very organic, designed in such a way that the different systems feed into one another. In a PvE-only game, focusing mainly on content, this isn’t a big deal. But in Pirates of the Burning Sea we have systems that require a minimum number of players to function correctly, such as our economy, and they break other systems if they’re not working correctly (such as PvP). If we didn’t have these kinds of interdependent systems, we wouldn’t even be considering server merges.
This is a measurement of how much time in a given day that the average player is in game. It was extremely high in the beta when everything was new, and then lowered once people settled in.
We’re not exactly sure what happened to our Spanish launch, but it didn’t go so well, considering that both Danish and Swedish players each outnumber Spanish players two to one. Not so good when you consider we dedicated an entire server to the Spanish language.
Pirates of the Burning Sea depends heavily on the conflict between the four nations, the British, Spanish, French and of course, Pirates. When these national populations are imbalanced, it’s not a problem when you’ve got high concurrency, because there are bottlenecks strategically placed in the Nation vs. Nation combat that enables a lower population nation to compete with a higher population nation. However, our target concurrency range was set without sufficient regard for this. We came to discover that when dealing with populations in the low end of the range, the problem wasn’t solved by the bottlenecks.
And the Rest
There are other factors, of course. But getting the above criteria wrong resulted in our concurrency targets being much lower than they should have been. Given that information the discussion becomes a focus on what you plan to do about it. For many MMOs, there’s a desire to maintain that feeling of positive momentum, to continue as though it is business as usual. Unfortunately, it often goes way too far. That overwhelming urge to avoid any sign of weakness creates crippling problems in these games in the long term. The company doesn’t want to admit it screwed up, and that prevents it from making the necessary changes, lest they be perceived as a sign of weakness. Which, in MMO terms, means you never shut down a server until its last gasps.
We could have gone that way. As multiple people have said in our forums, they felt like many of the servers we decided to merge were still perfectly viable. But as we found, the game works better with a higher concentration of players. So we decided to act.
There’s no better advertising than word of mouth. Over the next few days we’ll have stories about the servers merging, but what happens after that? It’s back to what actual players think and say. And if you make players happy, they tell their friends. That’s the real buzz, and that’s what we’re focused on.