[MUD-Dev] Architecture (Cell Rebalancing)

Daniel.Harman at barclayscapital.com Daniel.Harman at barclayscapital.com
Tue Jul 1 10:20:40 CEST 2003


From: Peter "Pietro" Rossmann [mailto:peter.rossmann at telia.com]

> I have started to think about the Cell Handler Rebalancing - the
> Cell Breathing.  If a cell detects that it is going to be
> overloaded, it should reshape itself.  But, given that the cell
> shape would be a rectangle, I can (and will) end up with pretty
> iregullar grid. Reshaping then could involve quite some number of
> neghbor cells on the same edge.

> Is there any better solution to this? As for example - triangulary
> shaped cells would have always only one neighbor on an edge? (if
> moving vertexes as if glued)

I'd argue that this approach is over engineered. If player density
is sufficiently high in an area to overload a properly scaled cell,
then in all likelihood the bandwidth & rendering requirements will
be so extreme that they'll dropping out anyway. Asheron's call used
to have a portal storm if too many congregated in an area and
randomly teleport people to another cell. Never liked it much, but
again it is a simpler approach.

There is an article about these movable cell boundaries in one of
the multiplayer game development books. It has some anecdotes about
how traumatic a scheme like this can be, and it was more than enough
to put me off.

Of course an alternative is to look at non-spacial paritioning. Do
cell's need to reflect geographical position? Its certainly not
required, but when you start factoring in AI navigation, object
interactions etc it seems that geographic association is needed at
least somewhere in the system.

Dan
_______________________________________________
MUD-Dev mailing list
MUD-Dev at kanga.nu
https://www.kanga.nu/lists/listinfo/mud-dev



More information about the mud-dev-archive mailing list