Dr Andrew C Aitchison wrote:
Has anyone come across X applications re-engineered for low bandwidth networks, This in the context of thin clients.

Low bandwidth or high latency ?

This is the usual response you can expect when asking this question. Bandwidth and latency are part of the same equation. Both bandwidth and latency are outstanding problems when running X applications over the Internet (and they are not the only problems, as there is still security, manageability, etc.).

To have a proof of that you would need a X application
that doesn't need roundtrips at all, so you would be
able to test the bandwidth variable in isolation. nxagent
is that application.

http://www.nomachine.com/documentation/building-components.php

Like Xnest, on which it is based, it is a X server for
its clients and a X client for the real server, but it is
a almost complete rewrite optimized to run as a component
of the NX proxy system.

nxagent eliminates all the roundtrips, except the very
few that are needed at its startup (not at startup of
any of its clients). Thus, if you think that only latency
is a problem, running nxagent you should have a clear idea
of the performance you will get when all the remaining X
applications in the world will be equally optimized for
high latency networks. Unfortunately if you try nxagent
over ssh -C, instead of running it through the usual NX
compression, you will find that bandwidth matters.

Having eliminated the roundtrips, X applications will
still be painfully slow. If you want to ensure really
good performance, you have to provide really good comp-
ression.

Enter NX now.

NX solves the latency problem. Once solved the latency
problem it treats bandwidth as part of the picture and
tries to achieve that good compression.

http://www.nomachine.com/documentation/NX-XProtocolCompression.php

The NX proxy system rewrites the wire X protocol to be
efficient on low bandwidth, high latency networks. It is
not a reengineering of the X protocol, because it uses the
the X protocol, but it is a layer managing the differences
between X over local socket communication and X over the
network.

> with this combination, a fast modem connection (say 33Kbps)
> isn't painful in terms of bandwidth.

Unfortunately with this combination X -is- painful,
otherwise Andrew C Aitchison would have not asked this
question ;-). Hope more people in the X community will
dedicate some time at trying NX in future. I really think
that it could unveil some misconception that are slowly
becoming facts given for assured.

/Gian Filippo.

_______________________________________________
Devel mailing list
[EMAIL PROTECTED]
http://XFree86.Org/mailman/listinfo/devel

Reply via email to