Re: Introductions

[madsdyd@challenge.dk]

On Tue, 9 Dec 2003, KANE Aaron wrote:

> This is a very important comment that I would like to be flush out on
> his list if possible.  Given the original proposal to have a large
> number of concurrent players, a distributed model was seen as very
> critical.  With that as a goal, a single server is not a pleasant
> option.  Encrypting network traffic seem the most realistic to me.

[...]

> Each players tree periodically does a sync as you correctly assumed.  
> No, it does works as described above.  What will happen is at the start
> of the game a "event" will be inserted in to the tree X seconds from the
> start.  When this event comes up the tree will sync and then push a new
> event onto the tree X seconds in the future to resync.  To allow for
> many many players I can't have a central server.  Each players tree runs
> the events on his own map, and his own units.  Only when a unit from
> player "A" walks onto a map space owned by player "B" is there any
> network traffic.  Each unit in that situation is required to post events
> to both trees. In this way you can have many many people playing all on
> one "big" world but only posting events to neighboring simulation trees
> when there is a need.

There is nothing stopping you from making a distributed server. It only 
requires that you _somehow_ can split your total game state into 
meaningful (gamewise) chunks and that these chunks can be synchronized in 
the border cases. In your case the obvious choice would seem to be the 
game world (splitting it in chunks)

I made a project about this a couple of years ago. The good thing about 
basing splits on "physical areas in the gameworld" is that the density 
seems to fix the problem with too many players in one arae. Like 
Tanenbaums explanation of not doing long-term schedulers for interactive 
systems: "If the system gets to many concurrent users, some will log off 
due to the bad performance". Likewise for a game: If 200 players crowds 
within the same game space, some will be killed/leave anyway.

Mads

-- 
Mads Bondo Dydensborg.                               madsdyd@challenge.dk
I disapprove of what you say, but I will defend to the death your right to
say it.
  - Beatrice Hall [pseudonym: S.G. Tallentyre], 1907 (many times wrongfully
    attributed to Voltaire)