massiv-devel Mailing List for Massively multiplayer game middleware

> I have read through the documentation, very impressive
> work the team has done!

Thank you.

> 1. I am curious, is there ongoing development?

All the major work finished at April 2004 when the project was completed and
successfully approved (it was a school project, you know). Several people
have been interested in using/improving the Massiv, mostly David Li, that
migrated the project from SF to ObjectWeb and wanted to estabilish a new
development team. Unfortunatelly we have not heard from him for quite a
while and do not know if the new team has made any progress.

> 2. If I were to use this in a commercial application,
> what kind of support is available?

I can help you understand the core internals. However I can not guarantee
anything.

Also there are some issues with the core that would probably have to be
resolved prior using the Massiv in a commerical application. These are
network subsystem, that now stupidly relies on TCP services provided by the
OS (we need a "session layer" that can overcome TCP connection dropping;
probably not to use TCP at all) and missing automatic archive management.
See TODO list on the web page for complete information.

On the other side, everything what we planned to support is fully
functional.

> 3. What kind of load can the system support?

We ran several clients (30 if I remember correctly) with disabled rendering
(ctrl+r) on a single box without problems. Also note that the current Demo
might be rather "slow" because it is designed to excersise the object model.
For example sectors were intentionally made very small (8x8 tiles; where
each entity occupies at least one tile) so that object migration and its
impact on the resource load could be tested (any inter-sector communication
is treated as if it were remote, because each sector can be on a different
server); there is a lot of small objects, everything is implemented as a
managed object (for example sector tile), etc. So certainly there is a
potential for a speedup. If sectors spanned hectars, the majority of the
load would be local. And yes, there is a load balancer.

> I am looking forward to getting into the guts of this
> system.  The most impressive part is the amount of
> documentation.  The reason most open source systems
> don't is their documentation.  Good job to all!

Thank you.
-- Markoid