New Technologies Attack the One-World Problem
Hugh Pickens writes "An MIT Technology Review article has new details on the challenges of a 'one world design' in Massively Multiplayer Online Games. Most games shard their servers, putting up artificial barriers between friends and family members. Technologies are now being developed to keep lots of players within a single world, some of them based off of the unique PvP-heavy title EVE Online. The best part - the technologies don't just apply to gaming. 'NASDAQ, for example, can be thought of as a very large MMO, supporting very large numbers of 'players' performing billions of transactions daily in a graphically intense environment, all within a single shard. Technologies that solve this problem effectively, says George Dolbier, technical lead for games and interactive entertainment at IBM, will have applications in any industry that requires spotting and reacting to trends, or "anything where behavior is dynamic and you need to move resources around rapidly."'"
WoW isn't the only MMO out there. Not all MMOs have numerous servers each supporting one single "world" with no way to move to the others. Guild Wars lets you play with anyone, anywhere, anytime in the world. Servers are specific to entire continents (with Asia fragmented into Japan, Taiwan, China, Korea), but people can group across those servers on a whim and play together.
Since GW is largely instance-based there might be fewer performance issues to deal with than with WoW in this regard, but my point is I just don't see how one would dedicate a "news" story to this topic when a popular title like GW has already tacked the issue.
Or did I totally misunderstand the topic altogether? It could be ^_^
I like basketball!!1!
The article seems to miss that EVE being space based allows a huge environment to be created easily, a few random number generators and a bit of tweaking and you have a whole universe of stars and asteroids and it's easily extendable after that.
The landscape in the likes of WoW is a lot more design intensive, you have features and locations with NPCs and dungeons and so on put in place. To double the population on the server you would need to either double the design/quest writing hours, add in a bunch of fractally generated landscape that would be relatively boring and largely pointless or go with the EQ2 route of opening up instances of zones which always seemed artifcial to me ("Hey are you in Common Lands 1,2,3,4 or 5?").
Shards are basically entire instances of the game world. They may (and often are) spread across multiple servers, and it is theoretically possible (although unwise) to run multiple shards on the same server. To make a suitably geeky reference for Slashdot, Shards are sort of like parallel game world universes, only your character only exists in one of the universes not in all of them.
I read the internet for the articles.
Although it is true that there is no end user experience of selecting a world, my guess is that it is still a shard based architecture based on location within the world. I base that guess on the observation that object rendering and latency seems to be dependent on the number of people and objects in an area. A densely crowded area has much more lag then a sparely populated area. It is not dependent, however, on how many users are currently logged in to the world.
It seems to me that SL is multi-shard but you don't explicitly select the shard. Moving from one part of the world to another may move you to a different shard.
You are correct to some degree, in that the Second Life 'Grid' is made up of about 10,000 'sims' each of which equates to a cpu and holds an area of land int he grid. Your avatar is always located in a sim, and you move between them when moving around the world. One of the issues is that a sim can currently hold about 50-70 avatars in one place (although you can hold a meeting at a corner and up this to 250 or so).
However in practice this is a relatively minor consideration. More specifically anyone inworld can talk to (IM) anyone else logged on at the same time, and all assets are held in one database, so neither people or possessions are sharded in any way whatsoever.
Lindon Lab have apparently declared that the current architecture should manage up to 100,000 concurrent connections. I'm not sure I believe that, but one of the reasons I've come around to thinking that Second Life in whatever form it evolves into over the next few years, will be the dominent virtuality going forward is because LL have faced and tackled the scalability issues so far with reasonable (not brilliant, but ok) success. There's more first mover advantage there than is commonly realized.