The Soapbox: The industry's obsession with shards

Sponsored Links

The Soapbox: The industry's obsession with shards
Disclaimer: The Soapbox column is entirely the opinion of this week's writer and does not necessarily reflect the views of Massively as a whole. If you're afraid of opinions other than your own, you might want to skip this column.

The MMO genre is now over a decade old, and in that time we've seen countless innovations in game design, graphics technology and hardware infrastructure. Some of these innovations have become so essential that without them a game looks cheap, old or backward. A functional market or auction mechanic now replaces the old meet-and-trade style barter of some early MMOs, for example, and an MMO without copious map or chat tools is seen as grossly incomplete. The limits of what is possible have been pushed gradually forward, and yet certain ideas that were formed in the genre's infancy still seem to stick to new titles like glue.

Sharded server models made a lot of sense in the early 2000s, when server hosting was expensive and the teams working on the server code were small. Those limitations have been rapidly shattered in recent years, but still new MMOs shard their communities into small groups. There are even alternative server models out there that are just as cost-effective as the sharded model but are devoid of the negative side-effects of smashing the community into hundreds of pieces.

Read on as I take a look at why developers rely on the sharded server model, the problems surrounding splitting communities and what alternative server models are out there.

Cost-effective hardware

Back when MMOs were just starting to emerge, the games industry was operating in a fledgling market and online games weren't seen as the money-makers they have now become. Today, MMOs make so much money that companies like CCP Games can throw $50,000 US at buying a single piece of server hardware. Combined with a dramatic drop in the price of computing over the past few years, the regular income a successful MMO collects gives its development studio an unprecedented ability to buy high-end server equipment.

Running a shard on higher-end machines will raise the limit on the number of players that shard can handle simultaneously, but at some point upgrades stop being a cost-effective way to squeeze more players into the game. The solution most developers come up with is simply to start more servers. By running separate copies of the server software on cheap hardware or generic server hosting, companies can support as many players as they want. Costs scale up linearly with the total number of players supported, and server population limits can be used to keep each server from exceeding the performance requirements of the hardware it's running on. From a business standpoint, it looks like a great solution.

Splitting communities

The problem with sharding isn't in the actual splitting of players across multiple servers. Any MMO with a large enough playerbase will eventually exceed the computational capacity that one server can provide, and splitting players across servers is a necessary step in balancing that load. The problem is with splitting the game into separate micro-communities, each with its own economics, politics and achievements. Players are restricted to interacting with a small subset of the overall game community.

This does allow more players and guilds to get "first" kills on raid bosses and lets several players worldwide have the same character name, but the strategy is not without its downfalls. The negative aspects of sharding an in-game community in this manner are far more detrimental than not being the first to kill a boss or get the name you want. If you find someone in real life who plays the same MMO as you, there's a high likelihood that you both play on different servers. If you ever want to play with your new friend, one of you will have to abandon your current friends and guild and pay for a server transfer or start a new character from scratch.

From a business perspective, I think companies should be trying to weigh up the additional revenue from occasional server transfers against the fact that having friends to play with in an MMO provides an extremely strong barrier to exit. If every player has the potential to play with every other player, it's much easier to make friends who may keep you in the game for years to come.

Alternative: EVE Online's single-shard model

EVE Online developer CCP Games has pioneered the single-shard server model since the game launched in 2003. It's not cheap or easy to keep all players in one game world; indeed, the cluster of servers EVE runs on is one of the most powerful supercomputers in the games industry. As the EVE population has grown over the years, developers have fought an ongoing battle to keep the server ahead of that expansion.

While the average MMO server houses only a few thousand players, the EVE supercluster is home to over 360,000 actively subscribed accounts. If you find out that a friend plays EVE, you don't have to worry about what server he's on and won't have to start a new character to play with him. Every player can interact with every other player, dramatically increasing opportunities for PvP and magnifying the scale of political or economic activities in the game.

If a major pirate organisation has locked down a system or other players are using up all the resources in your region, there is no other instance of the world you can swap to in order to avoid that competition. If a war is causing the price of minerals to skyrocket or if an alliance has just captured a station, there is no other copy of the game world in which that didn't happen. This lends actions in the game a very real, solid feel that isn't possible with a sharded MMO.

Alternative: RuneScape's free-move model

The standard sharded server model provides companies with a low-cost, scalable solution, but splitting the community has some negative effects on gameplay. EVE Online's single-shard model gives the game one massive cohesive community that isn't divided by artificial boundaries, but the technology required to make it work doesn't come cheap. The ideal solution would provide as many of the benefits of both systems with fewer of the drawbacks. Believe it or not, that solution has existed for over a decade with RuneScape's free-move server model.

RuneScape is a perfect example of an MMO that scales itself perfectly to match any size of community while keeping that community as one largely cohesive unit. There are hundreds of RuneScape servers, each supporting only a few thousand players at a time, but players are able to log into any server they want at any time. There are no server transfers, no miniature server communities and no boundaries between subscribing members. If you make a friend while playing RuneScape or meet someone in real life who plays, you're guaranteed to be able to play with that friend any time you want.

As each server only needs to support a few thousand players at a time, the cost of running RuneScape's servers is as low as 1/500th of the costs incurred running World of Warcraft. Population limits are also not necessary as players self-organise into the types of server they want. Players looking to avoid competition for resources like ore can find a server with just a few hundred players online, while those interested in trading would gravitate toward high-population servers. To help things along, Jagex designates certain servers as meeting points for trade or certain group activities.


Sometimes it feels as if almost every MMO development studio just defers to the choices made by the creators of other successful games without doing its own research. This is particularly visible when comparing server models, as most MMO coming out on the market are still using Ultima Online-style shards. I think we'll see more games released using a single-shard server structure in the coming years to create a cohesive community. RuneScape's free-move system or similar instanced-zone systems used in games like Champions Online are perfectly viable and scalable alternatives the standard sharding mechanism. I strongly believe that the industry should be adopting these models as standard practice and that it's time for shards to go the way of the dinosaurs.

Everyone has opinions, and The Soapbox is how we indulge ours. Join the Massively writers every Tuesday as we take turns atop our very own soapbox to deliver unfettered editorials a bit outside our normal purviews. Think we're spot on -- or out of our minds? Let us know in the comments!
All products recommended by Engadget are selected by our editorial team, independent of our parent company. Some of our stories include affiliate links. If you buy something through one of these links, we may earn an affiliate commission.
Popular on Engadget