diff options
Diffstat (limited to 'pipermail/nel/2000-December/000078.html')
| -rw-r--r-- | pipermail/nel/2000-December/000078.html | 80 |
1 files changed, 80 insertions, 0 deletions
diff --git a/pipermail/nel/2000-December/000078.html b/pipermail/nel/2000-December/000078.html new file mode 100644 index 00000000..e1cfefb6 --- /dev/null +++ b/pipermail/nel/2000-December/000078.html @@ -0,0 +1,80 @@ +<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN"> +<HTML> + <HEAD> + <TITLE> [Nel] Suggestion for the NeL network library / architecture</TITLE> + <LINK REL="Index" HREF="index.html" > + <LINK REL="made" HREF="mailto:archer%40nevrax.com"> + <LINK REL="Previous" HREF="000079.html"> + <LINK REL="Next" HREF="000080.html"> + </HEAD> + <BODY BGCOLOR="#ffffff"> + <H1>[Nel] Suggestion for the NeL network library / architecture</H1> + <B>Vincent Archer</B> + <A HREF="mailto:archer%40nevrax.com" + TITLE="[Nel] Suggestion for the NeL network library / architecture">archer@nevrax.com</A><BR> + <I>Tue, 12 Dec 2000 11:08:20 +0100</I> + <P><UL> + <LI> Previous message: <A HREF="000079.html">[Nel] Suggestion for the NeL network library / architecture</A></li> + <LI> Next message: <A HREF="000080.html">[Nel] Suggestion for the NeL network library / architecture</A></li> + <LI> <B>Messages sorted by:</B> + <a href="date.html#78">[ date ]</a> + <a href="thread.html#78">[ thread ]</a> + <a href="subject.html#78">[ subject ]</a> + <a href="author.html#78">[ author ]</a> + </LI> + </UL> + <HR> +<!--beginarticle--> +<PRE>According to Jean-Noel Moyne: +><i> You should have a look at PGM, which is an open reliable multicast +</I>><i> protocol designed by Cisco and TIBCO (the company I work for), which is +</I>><i> design to solve just this kind of problems in the most efficient way. +</I> +Hmmm, about everyone around started talking about, then dismissing it +as unsuited to our needs, but that's chiefly due to one misunderstanding. + +At first we didn't see the "server to server" sentence. + +We still don't know how well multicasting can be useful in our server +architecture, because so far, we have very few one-to-many messages +across servers. Most services discuss with a specific target on a specific +service. + +However, we still have lots of discussion for the "world service", +i.e. the service which "simulates" the world and run the various objects +and NPC agents. That one might be a candidate for a multicast protocol, +since we can expect replication of objects across multiple world services. + +This tie in with another question which I didn't had time to reply to, +regarding how to "split" the world across the multiple world services. +If the split is done by geography (the UO model), then there is little +need of a PGM-based protocol, because most objects will be accessed by +2 world services (edges), or at most 3 (T-intersections). However, if the +split is done instead by reference (i.e. we cluster an agent with the +agents that refer to it more often), then we have a higher redundancy +of agents on servers, and a multicast is probably the best method to +send message to these agents. + +-- +Vincent Archer Email: <A HREF="mailto:archer@nevrax.com">archer@nevrax.com</A> + +Nevrax France. Off on the yellow brick road we go! + +</pre> + + + +<!--endarticle--> + <HR> + <P><UL> + <!--threads--> + <LI> Previous message: <A HREF="000079.html">[Nel] Suggestion for the NeL network library / architecture</A></li> + <LI> Next message: <A HREF="000080.html">[Nel] Suggestion for the NeL network library / architecture</A></li> + <LI> <B>Messages sorted by:</B> + <a href="date.html#78">[ date ]</a> + <a href="thread.html#78">[ thread ]</a> + <a href="subject.html#78">[ subject ]</a> + <a href="author.html#78">[ author ]</a> + </LI> + </UL> +</body></html> |