cloud - What technologies exist for app level clustering? -

Assume I have an application that is guaranteed to drown a server even after optimization. Is there a technique that allows a web application to be split on multiple servers while maintaining a server?

For example, a multi-player online game Generally, in the craft of World War I have many servers with my game state which I see is something that this game state can do in many servers Will allow sharing.

EDIT: The platform does not really matter because it is for a web application, so there is no point in the backend. However, there are rail and Java possibilities.

Besides, Woah was just an example, I am interested in this kind of technology; A big game will be a good application for this.

In fact, for really low latency, high-throughput in-memory applications like real-time online, Not in the game, not as a piece of at least arbitrary middleware.

On the side of applicability and complexity, made a commendable effort, but found (surprisingly) that it was not scaling.

Eventually this is a difficult (though not rude) problem where there is no solution which is near the universal application. In particular, there is a likelihood of a correction between the need to act on stale game data rather than the need to continuously exchange data that is distributed among each other's hands. Lack of purity, exponential growth in complexity ... Pick your poison.

It's worth noting - especially if your application domain is not a real-time game - that you do not often mind if you work with stale data, until it gets right soon In such cases, simple caching systems like mammerskas are great. Similarly, if you need more precision but do not have to worry about throughput, then Hajjalk (mentioned in the second answer) Such can be very good, but for most online games that "thousands of Operations / sec" is not good enough.

Some MMO technology geographically makes some effort to distribute it by distributing the application, which means that there really is not a much more shared state, and this requires planning. Quote in the world and fantasy.

Another way is to divide it by service, and to implement most services from your favorite off-shelf RPC viewpoint. If your services are free, but any dependency between services puts you back into Square once, then it lets you easily scale.