1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
|
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<TITLE> [Nel] Re: Nel digest, Vol 1 #210 - 1 msg
</TITLE>
<LINK REL="Index" HREF="index.html" >
<LINK REL="made" HREF="mailto:nel%40nevrax.org">
<META NAME="robots" CONTENT="index,nofollow">
<LINK REL="Previous" HREF="000835.html">
</HEAD>
<BODY BGCOLOR="#ffffff">
<H1>[Nel] Re: Nel digest, Vol 1 #210 - 1 msg
</H1>
<B>nicO
</B>
<A HREF="mailto:nel%40nevrax.org"
TITLE="[Nel] Re: Nel digest, Vol 1 #210 - 1 msg">nel@nevrax.org
</A><BR>
<I>Wed, 26 Dec 2001 13:36:33 -0500</I>
<P><UL>
<LI> Previous message: <A HREF="000835.html">[Nel] Max Bezier Patches
</A></li>
<LI> <B>Messages sorted by:</B>
<a href="date.html#836">[ date ]</a>
<a href="thread.html#836">[ thread ]</a>
<a href="subject.html#836">[ subject ]</a>
<a href="author.html#836">[ author ]</a>
</LI>
</UL>
<HR>
<!--beginarticle-->
<PRE>I imagine that the data resend by the server are almost the same than
those send by the client. It isn't possible to mix this to technics ? We
can imagine peer to peer access for close characters to interract with a
minimuim latency. And for the farer charaters, we can use the server and
movement prediction (which could not be so bad because if there far, the
characters will be small on screen).
The idea behind that is to relax the pressure on the needed latency of
the server. Does it not possible to add some more intelligence inside
the client to predict what is usefull or not to minimize exchange ?
nicO
><i> Subject: RE: [Nel] peer to peer ?
</I>><i>
</I>><i> There are various reasons why we haven't opted for a peer to peer model.
</I>><i>
</I>><i> One of the most basic problems is this:
</I>><i> We are trying to achieve a high level of audio-visual immersion with scenes
</I>><i> in which tens or hundreds of individuals move about fluidly. The majority of
</I>><i> players only have 56k modems and bandwidth is a big issue. The problem with
</I>><i> peer to peer is that each player has to send and receive tens or hundreds of
</I>><i> packets per scene update at a rate of several scene updates per second. The
</I>><i> internet headers alone for these packets are liable to completely drown the
</I>><i> modem.
</I>><i>
</I>><i> Internally we use kunning systems to decide which information on which
</I>><i> players is relavent to send to which other players taking bandwidth
</I>><i> constraints into account. This allows us to make optimum use of the
</I>><i> bandwidth available.
</I>><i>
</I>><i> This code is tightly tied in with the game systems so I'm affraid it won't
</I>><i> be released for a while yet.
</I>><i>
</I>><i> Daniel
</I>><i>
</I>><i> -----Original Message-----
</I>><i> From: <A HREF="mailto:nel-admin@nevrax.org">nel-admin@nevrax.org</A> [mailto:<A HREF="mailto:nel-admin@nevrax.org">nel-admin@nevrax.org</A>]On Behalf Of
</I>><i> <A HREF="mailto:nicolas.boulay@ifrance.com">nicolas.boulay@ifrance.com</A>
</I>><i> Sent: Tuesday, December 18, 2001 1:16 PM
</I>><i> To: <A HREF="mailto:nel@nevrax.org">nel@nevrax.org</A>
</I>><i> Subject: [Nel] peer to peer ?
</I>><i>
</I>><i> -- __--__--
</I>><i>
</I>><i> Message: 3
</I>><i> From: "Vianney Lecroart" <<A HREF="mailto:lecroart@nevrax.com">lecroart@nevrax.com</A>>
</I>><i> To: <<A HREF="mailto:nel@nevrax.org">nel@nevrax.org</A>>
</I>><i> Subject: Re: [Nel] server architecture (front end,
</I>><i> load balancing, login service, ....)
</I>><i> Date: Mon, 17 Dec 2001 15:22:30 +0100
</I>><i> charset="iso-8859-1"
</I>><i> Reply-To: <A HREF="mailto:nel@nevrax.org">nel@nevrax.org</A>
</I>><i>
</I>><i> > But how clients are balanced between the N front
</I>><i> ends of a shard ?
</I>><i>
</I>><i> For now, the welcome service does a dumb balancement
</I>><i> (try to have the same
</I>><i> player number on each front end).
</I>><i>
</I>><i> >>>> Why don't you try to use LVS ?
</I>><i> More generaly do you plan to make direct connection
</I>><i> between client (such peer to peer network, or
</I>><i> distributed computing) ? It could be nice for
</I>><i> reduicing trafic to the server. To havoid cheating,
</I>><i> each client cheeck data as the server does. Maybe it
</I>><i> could be possible to connect people which interract a
</I>><i> lot (in the same area, ...), there is still some
</I>><i> mouvement with server to handle critical ressouces.
</I>><i> So mouvement to the center server are almost reduice
</I>><i> to the minimum.
</I>><i>
</I>><i> Does i smoke too much ?
</I>><i>
</I>><i> nicO
</I>><i>
</I>
</PRE>
<!--endarticle-->
<HR>
<P><UL>
<!--threads-->
<LI> Previous message: <A HREF="000835.html">[Nel] Max Bezier Patches
</A></li>
<LI> <B>Messages sorted by:</B>
<a href="date.html#836">[ date ]</a>
<a href="thread.html#836">[ thread ]</a>
<a href="subject.html#836">[ subject ]</a>
<a href="author.html#836">[ author ]</a>
</LI>
</UL>
</body></html>
|
