It just occurred to me that what could work is a compromise between a dedicated server-group and a peer to peer configuration:

Privately hosted dedicated servers. Each server would host a pre-defined, scalable set of resources, depending on the server''s capacity.

The very first server to go into operation would (obviously) have to support 100% of the game''s resources, including support for ALL clients (with a pre-defined limit, i.e. 5, 10, 50, or 100), all NPCs, and all environmental variables and terrain data.

Then when a second server comes online, it can be configured to provide support for the first, by taking over part of the job of the first and adding additional capacity, for instance. Naturally if the first goes offline the second takes 100% of the job of the first.

For every additional dedicated server that goes into service, the capacity is increased. Players can be kept aware of how many dedicated servers are in operation, and the approximate capacity of those servers.

Its no different really than a "server farm" in the sense that the operational servers share the workload in a relatively pre-defined manner.. but the obvious benefit is that the servers can be run by fans, cost little to nothing to play over time.

In fact, to further streamline the process (and justify some sort of minimal monthly income for the developer ), a server maintained by the developer/publisher could handle scheduling and dividing the workload amonst the player run servers.

It just occurred to me that what could work is a compromise between a dedicated server-group and a peer to peer configuration:

Privately hosted dedicated servers. Each server would host a pre-defined, scalable set of resources, depending on the server''s capacity.

The very first server to go into operation would (obviously) have to support 100% of the game''s resources, including support for ALL clients (with a pre-defined limit, i.e. 5, 10, 50, or 100), all NPCs, and all environmental variables and terrain data.

Then when a second server comes online, it can be configured to provide support for the first, by taking over part of the job of the first and adding additional capacity, for instance. Naturally if the first goes offline the second takes 100% of the job of the first.

For every additional dedicated server that goes into service, the capacity is increased. Players can be kept aware of how many dedicated servers are in operation, and the approximate capacity of those servers.

Its no different really than a "server farm" in the sense that the operational servers share the workload in a relatively pre-defined manner.. but the obvious benefit is that the servers can be run by fans, cost little to nothing to play over time.

In fact, to further streamline the process (and justify some sort of minimal monthly income for the developer ), a server maintained by the developer/publisher could handle scheduling and dividing the workload amonst the player run servers.

****************************************

Brian Lacy
ForeverDream Studios

Comments? Questions? Curious?
brian@foreverdreamstudios.com

"I create. Therefore I am."

There are two problems (off the top of my head) with P2P server processing:

1) If Kazaa is any indication, people won''t want to share. I share anything I DL off Kazaa, and sign on to Kazaa when my computer is idle, so that people can download from me. You would need to give players a reason to share their CPU power.

2) A lot of people still have dial-up connections, and could do only a minimal amount of work. I think this is less of an issue, though, because most people (I think) who play games are going to have a high speed connection. However, I have what''s called ADSL, which means I have a slow upload speed; I''m not sure how useful I could be as part of a P2P processing network.

quote: Original post by Peon
You would need to give players a reason to share their CPU power.

You mean like having to share processsor power in order to play the game?

Connection speed is still an unsolved problem though...