Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!cs.utexas.edu!csd4.milw.wisc.edu!lll-winken!lll-crg.llnl.gov!casey
From: casey@lll-crg.llnl.gov (Casey Leedom)
Newsgroups: comp.windows.x
Subject: DRAFT PROPOSAL: Proxy Server / Workstation Agent Protocol
Message-ID: <26683@lll-winken.LLNL.GOV>
Date: 9 Jun 89 21:50:20 GMT
Sender: usenet@lll-winken.LLNL.GOV
Reply-To: casey@lll-crg.llnl.gov (Casey Leedom)
Organization: Lawrence Livermore National Laboratory
Lines: 62


| From: lupine!ed@uunet.UU.NET (Ed Basart)
| Date: Wed, 7 Jun 89 08:24:30 PDT
|
|	Ed Basart, NCD, Inc.
|	(415)694-0650
|	ed@ncd.com
| 
| We here at NCD are quite interested in the problem of running X over a
| dial-up modem.  However our point of departure is that the server runs in
| the X terminal.  The reason for this is adherence to the X architecture -
| separate the problem at the client/server boundary.  It also makes good
| sense.  X running on a Sun 3/50 pretty much consumes the entire processor
| when drawing to the screen.  Trying to run multiple at-home X servers on
| the host will create performance problems and herky-jerky response.  This
| can be demonstrated by loading a few X servers on a diskful Sun, and then
| try pounding away.


  Unfortunately I can't agree with you.  And several people who've run
the Graphon Optimax 200 say that the Proxy Window Servers don't seem to
wipe out the host that badly.  You've got to remember where a lot of
those CPU cycles are burned in an X11 server: in the low level graphics
code.  The design I'm proposing still has that code on the Workstation.
And I think the problems with server memory are abundantly clear.

  I think that it's going to be far more cost effective to use
already in place equipment like personal computers to run window systems
on.  And for many of those systems, putting a full X11 Window Server on
them with a full TCP/IP SLIP networking system is just more work than you
want to think about.

  For those who don't have and don't want a personal computer, then
there's something like an X terminal.  I see my proposal as setting up
the possibility of extremely cheap, cost effective, medium performance
Workstations that a company like yours could manufacture.  It's quite
possible that the cost could get as low as $500.

| Regardless, we already have a product that runs the server only the
| the terminal (we prefer to call it a station), and now are writing
| the software to support our station across a dial-up line.  Our
| experiments have shown SLIP to be OK at 38.4 Kbits, tolerable at
| 19.2 Kbits, and unacceptable at 9.6Kbits.  Our initial objective
| is rather modest:  a 4 to 1 improvement over SLIP.  Merely removing
| the TCP headers may be enough.  We are now doing an implementation
| of an X serial protocol server (daemon) and client on the Sun to
| try our ideas out.
| 
| If we get the X serial protocol to work, we'll gladly push it as
| a standard.  We have been waiting around for "someone else" to
| do it, and will give it a go ourselves.

  You probably could get some form of compressed X protocol to work and
be fairly bandwidth efficient, but then you'd be limiting yourself to X.
``What?'' you're saying, ``limiting myself to X???'' And I'd have to
answer that my first proposal did the same thing, but a kind soul pointed
out the errors of my ways.  I'll send a copy of my current proposal in a
follow up.  It should also be pointed out that this will require a Proxy
Server for the protocol conversion, but I also cover this in my proposal
...

Casey