Re: Mesh linking - technical (Was: Re: [Unreal-users] Planned features for 3.3?)

[christian s. <han...@gm...>]

Just to cast my .02 into the ring..here are my thoughts.

There is just really no way that you can (sanely) do full mesh All:All
linking on larger networks.
That said, here are some of my thoughts on the idea.

In order to handle desynch issues, a simple syn/ack type system could
be implemented.

User on Server A sets mode +i
Server A sends mode change to Servers B and C
Server B acks, Server C does not
Server A tries n more times to update Server C
(now let's just get silly...)
Server A passes a flag around the mesh claiming Server C is unreachable
Other servers in mesh try a test syn/ack, No response from Server C!
Servers Auto-Jupe Server C and an event is logged/triggered

That's at least one way it could be done (the Jupeing is a bit silly
really). Another idea I could see scaling a little better is this:
When a server joins the "mesh" it sends a ping type request, and based
on the latency/response time of the ping-replys it assigns each server
in the mesh it's own internal metric. Any server over a certain metric
is dropped from it's internal table of "Servers to send updates to".

The idea being that rathers than sending updates to *every* server, we
will instead send updates to every *close* (in terms of speed, not
physical location, obviously) Server. The nice thing about this is
that we could hard assign a metric. For example, we have a server in
NY and a server in LA, we realize that there are times when these two
servers may not respond fast enough for the cut, but we also know that
these are both high traffic primary servers on our network, so we hard
code a low metric so that updates are always sent from to the other,
and to help eliminate desync, we also use the syn/ack system (sans
silly jupe idea).

--=20
Christian Sullivan

One Pill Makes You Larger,
and one pill makes you small.

http://freezerpants.com ||  http://superimposable.org