Speed up or replace the vnc repeater

[filippo]

Hi all,

I have setup a system of two ultraVnc clients (behind symmetrical NATs) and my own vnc repeater(on a public IP) for testing. I have been using ultraVNC 1.0.5 on XP and Vista machines, repeater rel1.0 on a window server 2003.

The system is working but I experience the result to be very slow when using the repeater.

How can I make the repeater work better and faster?

I have an idea of using an implementation of a ICE server to replace the vnc repeater and change the ultraVnc server/viewer source code to work with my implementation. This will require some coding efforts from me and I'm not sure if this actually will improve the performance. Any thoughts on this?

What I want to achieve is a feature for desktop sharing over the Internet which can be intregrated into a larger project. I have access to both a windows system and a linux system with public IPs.

All ideas on how I can improve the vnc repeater or make a better solution for relaying will be greatly appreciated...

[mod=494,1240868864]moved from 105.6 to SC[/mod]

[MrScotty]

I had to look at the repeater (and some port forwarding tools like NetworkActiv AUTAPF ).
I've found that when connecting directly(viewer to server) everything is working fast and smooth. When using the repeater things becomes slower.
In my test i use a generator to make controlled screen updates. So i have a square of 512x512 changing every 100 mSecs (10 fps).
When connected through the repeater i noticed that the updates stops/become slow when i move the mouse.
I performed a test where I used two socket connection (port 5900 and 5901) where i used one to send to the server and the other to receive messages from the server.
Using it directly it was fast using it with AUTAPF it was slow.
I mesaured the time between two rfb update requests and found it was 200 mSec (DingDong ). 200 mSecs sounds like Delayed ACK and related to it the Nagle Algorithm.
Next i adjusted the repeater code to disable the Nagle algortihm and performed the test again. Things where performing as expected. No hickups in the updates when moving the mouse around.

In short to get better repeater performance, I think, disabling Nagle's algoritme should improve the speed.

[JDaus]

In short to get better repeater performance, I think, disabling Nagle's algoritme should improve the speed.

Its about time sombody paid some attention to the old repeater ... its been lagging behind for some years now ...

I used it once upon a time, but now stick with SSH ...

'If you build it, they will come'

[redge]

I agree for remove the Nagle's algoritme of repeater as you said Mr Scotty

[redge]

MrScotty

beta repeater 1.0.8.x Nagle's algoritm disabled
[topic=16510][/topic]

tested on LAN OK