Relative speeds of ways of doing 'X over the internet'?
Does anyone have any ideas about, or is there a web site that will tell me about, the relative speeds of different ways of running X applications across the internet? I'm specifically interested in runnning jpilot (a Palm Dekstop replacement that runs on Linux) remotely through my ADSL connection. The only way I have run it to date is tunnelling X through an ssh connection. This works OK in the sense that it's full functional but it's not really fast enough for normal use. It's just a way of getting at the data if I really need to. What I'm after is knowing if there are better and/or more optimised ways of connecting so that this would become a usable utility when accessed remotely. In terms of bandwidth use jpilot is sort of 'middling', it is much, much more usable than running Firefox via the ssh tunnel for example but not as quick as running an rxvt terminal. (An rxvt terminal actually runs quite quickly but there's little point as it only offers the same function as an ordinary ssh session which is quicker and easier) So, can anyone offer any ideas please. -- Chris Green (chrisisbd@leary.csoft.net)
On 25-May-06 chrisisbd@leary.csoft.net wrote:
Does anyone have any ideas about, or is there a web site that will tell me about, the relative speeds of different ways of running X applications across the internet? I'm specifically interested in runnning jpilot (a Palm Dekstop replacement that runs on Linux) remotely through my ADSL connection.
The only way I have run it to date is tunnelling X through an ssh connection. This works OK in the sense that it's full functional but it's not really fast enough for normal use. It's just a way of getting at the data if I really need to.
What I'm after is knowing if there are better and/or more optimised ways of connecting so that this would become a usable utility when accessed remotely.
There are a couple of possibilities you might consider: -- LBX (Low Bandwidth X) -- dxpc - (Differential X Protocol Compressor) which are designed to speed up X over remote connections. They are outlined in http://www.paulandlesley.org/faqs/LBX-HOWTO.html I have used LBX in the past over a dialup link, but it was years ago. Though it's not mentioned in the above description, I was under the impression (even for LBX) that there was also an element of not sending redundant information -- only what is needed to implement changes in the display. Both of the above are, I think, rather old. Whether they're still available for modern X I can't say. Good luck! Ted. -------------------------------------------------------------------- E-Mail: (Ted Harding) <Ted.Harding@nessie.mcc.ac.uk> Fax-to-email: +44 (0)870 094 0861 Date: 25-May-06 Time: 09:56:26 ------------------------------ XFMail ------------------------------
On Thu, May 25, 2006 at 09:56:30AM +0100, Ted Harding wrote:
On 25-May-06 chrisisbd@leary.csoft.net wrote:
Does anyone have any ideas about, or is there a web site that will tell me about, the relative speeds of different ways of running X applications across the internet? I'm specifically interested in runnning jpilot (a Palm Dekstop replacement that runs on Linux) remotely through my ADSL connection.
The only way I have run it to date is tunnelling X through an ssh connection. This works OK in the sense that it's full functional but it's not really fast enough for normal use. It's just a way of getting at the data if I really need to.
What I'm after is knowing if there are better and/or more optimised ways of connecting so that this would become a usable utility when accessed remotely.
There are a couple of possibilities you might consider:
-- LBX (Low Bandwidth X) -- dxpc - (Differential X Protocol Compressor)
which are designed to speed up X over remote connections. They are outlined in
http://www.paulandlesley.org/faqs/LBX-HOWTO.html
I have used LBX in the past over a dialup link, but it was years ago.
Though it's not mentioned in the above description, I was under the impression (even for LBX) that there was also an element of not sending redundant information -- only what is needed to implement changes in the display.
Both of the above are, I think, rather old. Whether they're still available for modern X I can't say.
They sound much more the sort of thing I'm after, thanks. As you say though they are rather old. I'll have a good hunt around with Google and see what I can find out. -- Chris Green (chrisisbd@leary.csoft.net)
On 5/25/06, chrisisbd@leary.csoft.net <chrisisbd@leary.csoft.net> wrote:
Does anyone have any ideas about, or is there a web site that will tell me about, the relative speeds of different ways of running X applications across the internet?
Investigate VNC - it was designed for this. Hope this helps, Tim.
On Thu, May 25, 2006 at 11:27:14AM +0100, Tim Green wrote:
On 5/25/06, chrisisbd@leary.csoft.net <chrisisbd@leary.csoft.net> wrote:
Does anyone have any ideas about, or is there a web site that will tell me about, the relative speeds of different ways of running X applications across the internet?
Investigate VNC - it was designed for this.
My experience of VNC is that it is actually rather slower than using X directly. VNC also is overkill for what I want to do, as far as I understand it VNC always gives you the complete desktop from the remote (server) machine, I just want a single application. -- Chris Green (chrisisbd@leary.csoft.net)
Tim Green wrote:
On 5/25/06, chrisisbd@leary.csoft.net <chrisisbd@leary.csoft.net> wrote:
Does anyone have any ideas about, or is there a web site that will tell me about, the relative speeds of different ways of running X applications across the internet?
Investigate VNC - it was designed for this.
Hope this helps, Tim.
I use the free version of VNC for things like this and use custom port numbers which make it a little harder to sniff. I only use it in small bursts and I ensure that I never type passwords or enter user ID's into web browsers whilst using the connection. Keystrokes other than the initial password request during connection, are unencrypted. Real VNC free edition can be used with speed and safety if used responsibly, thats my opinion.
On Thu, May 25, 2006 at 11:46:10AM +0100, David Simon Cooper wrote:
Tim Green wrote:
On 5/25/06, chrisisbd@leary.csoft.net <chrisisbd@leary.csoft.net> wrote:
Does anyone have any ideas about, or is there a web site that will tell me about, the relative speeds of different ways of running X applications across the internet?
Investigate VNC - it was designed for this.
Hope this helps, Tim.
I use the free version of VNC for things like this and use custom port numbers which make it a little harder to sniff. I only use it in small bursts and I ensure that I never type passwords or enter user ID's into web browsers whilst using the connection. Keystrokes other than the initial password request during connection, are unencrypted.
Err, I think you'll find the initial password also travels over the net unencrypted. It certainly used to. Cheers, -- Brett Parker
On 5/25/06, Brett Parker <iDunno@sommitrealweird.co.uk> wrote:
On Thu, May 25, 2006 at 11:46:10AM +0100, David Simon Cooper wrote:
Tim Green wrote:
On 5/25/06, chrisisbd@leary.csoft.net <chrisisbd@leary.csoft.net> wrote:
Does anyone have any ideas about, or is there a web site that will tell me about, the relative speeds of different ways of running X applications across the internet?
Investigate VNC - it was designed for this.
I use the free version of VNC for things like this and use custom port numbers which make it a little harder to sniff. I only use it in small bursts and I ensure that I never type passwords or enter user ID's into web browsers whilst using the connection. Keystrokes other than the initial password request during connection, are unencrypted.
Err, I think you'll find the initial password also travels over the net unencrypted. It certainly used to.
In which case tunnel it over ssh or a vpn. Tim.
On Thu, May 25, 2006 at 12:08:15PM +0100, Tim Green wrote:
On 5/25/06, Brett Parker <iDunno@sommitrealweird.co.uk> wrote:
On Thu, May 25, 2006 at 11:46:10AM +0100, David Simon Cooper wrote:
Tim Green wrote:
On 5/25/06, chrisisbd@leary.csoft.net <chrisisbd@leary.csoft.net> wrote:
Does anyone have any ideas about, or is there a web site that will tell me about, the relative speeds of different ways of running X applications across the internet?
Investigate VNC - it was designed for this.
I use the free version of VNC for things like this and use custom port numbers which make it a little harder to sniff. I only use it in small bursts and I ensure that I never type passwords or enter user ID's into web browsers whilst using the connection. Keystrokes other than the initial password request during connection, are unencrypted.
Err, I think you'll find the initial password also travels over the net unencrypted. It certainly used to.
In which case tunnel it over ssh or a vpn.
*All* of which is entirely irrelevant to my original question! :-) Running application xxx directly as an X application using an ssh tunnel will surely be faster than running application xxx on the remote system and then running the whole X desktop via VNC (over an ssh tunnel or not). OK, it's not so portable as I need to be running an X server at the 'client' end (where I want to sit and use the application) but that's my situation. I want to run a remote X client application on my X desktop, it's what X is designed for. All I want are ways to optimise the X traffic across the internet, if that's possible. ... and no, it's *not* true that VNC was "designed for this", VNC was designed for remote control of another computer and that is most definitely not what I want to do. It could well be that someone else is using that remote computer at the remote location. My X application will run quite happily without them even noticing (unless I hog lots of processor of course). -- Chris Green (chrisisbd@leary.csoft.net)
chrisisbd@leary.csoft.net wrote:
On Thu, May 25, 2006 at 12:08:15PM +0100, Tim Green wrote:
On 5/25/06, Brett Parker <iDunno@sommitrealweird.co.uk> wrote:
On Thu, May 25, 2006 at 11:46:10AM +0100, David Simon Cooper wrote:
Tim Green wrote:
On 5/25/06, chrisisbd@leary.csoft.net <chrisisbd@leary.csoft.net> wrote:
Does anyone have any ideas about, or is there a web site that will tell me about, the relative speeds of different ways of running X applications across the internet?
Investigate VNC - it was designed for this.
I use the free version of VNC for things like this and use custom port numbers which make it a little harder to sniff. I only use it in small bursts and I ensure that I never type passwords or enter user ID's into web browsers whilst using the connection. Keystrokes other than the initial password request during connection, are unencrypted.
Err, I think you'll find the initial password also travels over the net unencrypted. It certainly used to.
In which case tunnel it over ssh or a vpn.
*All* of which is entirely irrelevant to my original question! :-)
Running application xxx directly as an X application using an ssh tunnel will surely be faster than running application xxx on the remote system and then running the whole X desktop via VNC (over an ssh tunnel or not).
OK, it's not so portable as I need to be running an X server at the 'client' end (where I want to sit and use the application) but that's my situation. I want to run a remote X client application on my X desktop, it's what X is designed for. All I want are ways to optimise the X traffic across the internet, if that's possible.
... and no, it's *not* true that VNC was "designed for this", VNC was designed for remote control of another computer and that is most definitely not what I want to do. It could well be that someone else is using that remote computer at the remote location. My X application will run quite happily without them even noticing (unless I hog lots of processor of course).
I think the commercial version of NX offers this ability ( Not tried) Full desktop or individual apps can be accessed http://www.nomachine.com/companion_screenshots.php We use freeNX to allow kids to access their school desktops from home. It runs over ssh and uses the square root of nought in bandwidth terms -- Regards Andy Trevor Technical Director Tel: 07840 988311 Cutter Project Limited http://www.cutterproject.co.uk -- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean.
On Thu, May 25, 2006 at 01:06:41PM +0100, Brett Parker wrote:
On Thu, May 25, 2006 at 06:41:42AM -0500, chrisisbd@leary.csoft.net wrote:
*All* of which is entirely irrelevant to my original question! :-)
Don't be so picky ;)
I assume that with ssh -X you're also using -C for the value added compression factor?
Yes, I tried adding the -C but to be quite honest it didn't seem to make a *lot* of difference though it ran a bit faster. -- Chris Green (chrisisbd@leary.csoft.net)
chrisisbd@leary.csoft.net wrote:
On Thu, May 25, 2006 at 12:08:15PM +0100, Tim Green wrote:
On 5/25/06, Brett Parker <iDunno@sommitrealweird.co.uk> wrote:
On Thu, May 25, 2006 at 11:46:10AM +0100, David Simon Cooper wrote:
Tim Green wrote:
On 5/25/06, chrisisbd@leary.csoft.net <chrisisbd@leary.csoft.net> wrote:
Does anyone have any ideas about, or is there a web site that will tell me about, the relative speeds of different ways of running X applications across the internet?
Investigate VNC - it was designed for this.
I use the free version of VNC for things like this and use custom port numbers which make it a little harder to sniff.
*All* of which is entirely irrelevant to my original question! :-)
Running application xxx directly as an X application using an ssh tunnel will surely be faster than running application xxx on the remote system and then running the whole X desktop via VNC (over an ssh tunnel or not).
I found using TightVNC (which adds compression) with a minimal desktop was faster than using remote X with ssh (even with compression turned on).
... and no, it's *not* true that VNC was "designed for this", VNC was designed for remote control of another computer and that is most definitely not what I want to do. It could well be that someone else is using that remote computer at the remote location.
You can start the vncserver up as it's own xserver that is only visible when vnc'ing to the machine and can exist at the same time as the local xserver that is shown on the monitor. I used to use exactly this setup to run a single application. I had a minimal window manager in the vncserver and I ran a single app. I could connect and disconnect as I wanted and the app would stay running and none of it interfered with the normal xserver. Of course now I think I might try NX rather than TightVNC but on my last attempt I failed to make it work properly. JD
On Thu, May 25, 2006 at 01:18:12PM +0100, Jon Dye wrote:
chrisisbd@leary.csoft.net wrote:
On Thu, May 25, 2006 at 12:08:15PM +0100, Tim Green wrote:
On 5/25/06, Brett Parker <iDunno@sommitrealweird.co.uk> wrote:
On Thu, May 25, 2006 at 11:46:10AM +0100, David Simon Cooper wrote:
Tim Green wrote:
On 5/25/06, chrisisbd@leary.csoft.net <chrisisbd@leary.csoft.net> wrote: > Does anyone have any ideas about, or is there a web site that will > tell me about, the relative speeds of different ways of running X > applications across the internet?
Investigate VNC - it was designed for this.
I use the free version of VNC for things like this and use custom port numbers which make it a little harder to sniff.
*All* of which is entirely irrelevant to my original question! :-)
Running application xxx directly as an X application using an ssh tunnel will surely be faster than running application xxx on the remote system and then running the whole X desktop via VNC (over an ssh tunnel or not).
I found using TightVNC (which adds compression) with a minimal desktop was faster than using remote X with ssh (even with compression turned on).
... and no, it's *not* true that VNC was "designed for this", VNC was designed for remote control of another computer and that is most definitely not what I want to do. It could well be that someone else is using that remote computer at the remote location.
You can start the vncserver up as it's own xserver that is only visible when vnc'ing to the machine and can exist at the same time as the local xserver that is shown on the monitor.
I used to use exactly this setup to run a single application. I had a minimal window manager in the vncserver and I ran a single app. I could connect and disconnect as I wanted and the app would stay running and none of it interfered with the normal xserver. Of course now I think I might try NX rather than TightVNC but on my last attempt I failed to make it work properly.
I might try tightvnc then and see how it goes, thanks. -- Chris Green (chrisisbd@leary.csoft.net)
Brett Parker wrote:
On Thu, May 25, 2006 at 11:46:10AM +0100, David Simon Cooper wrote:
Err, I think you'll find the initial password also travels over the net unencrypted. It certainly used to.
Cheers,
Hey Brett, from Real VNC "VNC Free Edition and older VNC 3 based systems support a simple challenge-response protocol used to verify a password of up to eight characters, supplied by the connecting user. While this avoids exposing the password to attackers as would be the case with pure plaintext protocols such as telnet, the rest of the session is unencrypted and so anything typed into the viewer passes "in the clear" to the server. VNC Free Edition is therefore suitable for use within a local network or secure VPN, but not for general use over untrusted networks, such as the Internet." They don't promote the use over the internet, but then we know that. I am wondering, when they say challenge-response do you think it means the response is not exncrypted? It would be pretty pointless not sending the password in the challenge incase it gets intercepted but all password tries in the response are left as plain text? I am assuming the response password is sent as some kind of encrypted value rather than just characters.
On Thu, May 25, 2006 at 01:27:58PM +0100, David Simon Cooper wrote:
They don't promote the use over the internet, but then we know that. I am wondering, when they say challenge-response do you think it means the response is not exncrypted? It would be pretty pointless not sending the password in the challenge incase it gets intercepted but all password tries in the response are left as plain text?
I am assuming the response password is sent as some kind of encrypted value rather than just characters.
Generally, what happens with challenge response is that the server sends a "challenge", so potentially an encryption method, a salt, or something similar and the client then sends a response encoded using the challenge. So, in a unix type system the challenge could, I suspect, be a 2 char password (DES), and the response be the password crypted with the salt (at least, that's my basic understanding of challenge-response). Thanks, -- Brett Parker
If we are talking about just jpilot, you could always just copy the $HOME/.jpilot directories using scp or rsync. Have you thought about getting broadband, Latency is a pain but its bandwidth that really counts for jpilot usage as I have used ssh tunnelled X11 applications from the UK at many sites in Europe. If you have a laptop you could use the free wireless broadband at the east oxford community centre. Have you selected a gtk theme with little colour? Have you considered using just 8 bits per pixel colour depth, Also have you thought of recompiling gtk as I believe toolkit has some support (like motif) for passing only the API calls across the network rather than the X11 commands. And finally have you thought about syncing your palm more frequently? I have found palm to be very good at syncing with multiple jpilots. Regards Owen Synge On Thu, 25 May 2006 05:08:49 -0500 chrisisbd@leary.csoft.net wrote:
Does anyone have any ideas about, or is there a web site that will tell me about, the relative speeds of different ways of running X applications across the internet? I'm specifically interested in runnning jpilot (a Palm Dekstop replacement that runs on Linux) remotely through my ADSL connection.
The only way I have run it to date is tunnelling X through an ssh connection. This works OK in the sense that it's full functional but it's not really fast enough for normal use. It's just a way of getting at the data if I really need to.
What I'm after is knowing if there are better and/or more optimised ways of connecting so that this would become a usable utility when accessed remotely.
In terms of bandwidth use jpilot is sort of 'middling', it is much, much more usable than running Firefox via the ssh tunnel for example but not as quick as running an rxvt terminal. (An rxvt terminal actually runs quite quickly but there's little point as it only offers the same function as an ordinary ssh session which is quicker and easier)
So, can anyone offer any ideas please.
-- Chris Green (chrisisbd@leary.csoft.net)
_______________________________________________ main@lists.alug.org.uk http://www.alug.org.uk/ http://lists.alug.org.uk/mailman/listinfo/main Unsubscribe? See message headers or the web site above!
On Sat, May 27, 2006 at 09:01:00AM +0100, Owen Synge wrote:
If we are talking about just jpilot, you could always just copy the $HOME/.jpilot directories using scp or rsync.
That's a thought that I had while lying in bed this morning and I think it may well satisfy my needs. I can schedule rsync to copy to and from my work system at appropriate times.
Have you thought about getting broadband, Latency is a pain but its bandwidth that really counts for jpilot usage as I have used ssh tunnelled X11 applications from the UK at many sites in Europe. If you have a laptop you could use the free wireless broadband at the east oxford community centre.
I have broadband! I still don't find that jPilot is really usable.
Have you selected a gtk theme with little colour? Have you considered using just 8 bits per pixel colour depth,
Also have you thought of recompiling gtk as I believe toolkit has some support (like motif) for passing only the API calls across the network rather than the X11 commands.
I'm not all that flexible at the remote end as it's a Sun Solaris system. Which 'end' affects this when selecting the gtk theme? (By remote end I mean the place where I want to use jPilot). -- Chris Green (chrisisbd@leary.csoft.net)
On Sat, 27 May 2006 06:12:11 -0500 chrisisbd@leary.csoft.net wrote:
On Sat, May 27, 2006 at 09:01:00AM +0100, Owen Synge wrote:
If we are talking about just jpilot, you could always just copy the $HOME/.jpilot directories using scp or rsync.
That's a thought that I had while lying in bed this morning and I think it may well satisfy my needs. I can schedule rsync to copy to and from my work system at appropriate times.
Also have you thought of recompiling gtk as I believe toolkit has some support (like motif) for passing only the API calls across the network rather than the X11 commands.
I'm not all that flexible at the remote end as it's a Sun Solaris system. Which 'end' affects this when selecting the gtk theme? (By remote end I mean the place where I want to use jPilot).
I think it depends whether gtk is aware that it is runnning remotely, in which case things are optimal, and it will look the same as your local applications, and if its not it will be in the style of the remote host. I am one of the people who thinks that remote X11 is a very good, particularly for abstraction. By the way its only second hand the Gtk can do this, I know Motif can. Regards Owen
participants (8)
-
Andy Trevor -
Brett Parker -
chrisisbd＠leary.csoft.net -
David Simon Cooper -
Jon Dye -
Owen Synge -
Ted.Harding＠nessie.mcc.ac.uk -
Tim Green