NX - A Revolution In Network Computing?

Anonymous Coward writes "Judging from this interview, it looks like KDE developers have found a new toy to add to their desktop's networking capabilities. They claim to be able to cram a fullscreen KDE session -- KMail for mailing, Konqueror for file management, Mozilla for web browsing and OpenOffice for word processing -- into a 40 KBit/sec modem connection without losing responsiveness for the user experience. At aKademy, the 9 day KDE Community World Summit, a group of core developers started to work on NX/FreeNX integration to help facilitate the "re-invention of the KDE desktop environment" for KDE4. Knoppix-3.6 is the first Linux distribution to ship an integrated FreeNX server (created by Fabian Franz) with the NoMachine NX Client."

Comments

Educate me.

[Limburgher]

Is this on top of a remote X display, or in place of one?

Re:Educate me.

[merlin_jim]

It seems to be an alternate way to pass messages back and forth.

I don't know if you've noticed but remoteX doesn't work in a bandwidth constrained environment. They're claiming that this does.

I think the app has to be NX-aware for it to work, however...

Re:Educate me.

[hummassa]

It's a remote graphics protocol based on X11, but with considerably less round-trips, due to aggressive caching on both sides.

Trying to be more plain:

imagine the work when you click on a button (exaggerated):

server: move your mouse position to x1, y1
client: move your mouse cursor to x1, y1
.
.
.
server: move your mouse position to x2, y2
client: move your mouse cursor to x2, y2; highlight button(button 1)
.
.
.
server: move your mouse position to x3, y3;
client: move your mouse cursor to x3, y3;
server: mouse down
client: display pressed button
server: mouse up
client: display pressed button (client will now do the ON_CLICK event)

under NX:

server: button(button 1) was clicked
client: does the ON_CLICK event

Re:Educate me.

[ahfoo]

I think the app has to be NX-aware for it to work, however...

I don't think you're right on that point. I downloaded the Knoppix3.6 iso with Bittorrent almost a week ago and I've been using Fabian's NX server the whole time since then. It gives you everything you get in a regular Knoppix KDE desktop. You can burn DVDs using K3B from a machine in another room among other things I've been doing lately.
I just wish there was some way to make it work at boot time so I could ditch my KVMs.
I did see that small /. thread on hardware IP KVMs the other day though. Sounded great, but I don't have one to play with. But hey if KVM over IP works for hardware, why not software. Sounds crazy, but you never know.

Re:Educate me.

[ADRA]

Two words: BIOS Support

In order to support boot style KVM abilities that hardware KVM's offer, the BIOS has to boot an entire IP stack and daemon to support the integrated KVM functionality, and don't forget the network drivers needed to support the network card. This would be quite heavy. It isn't as simple as the interface between the bios and net boot roms.

I guess this could actually be integrated into network boot roms and add a new differentiator to the innovation stagnent NIC ASIC market. Just to note, the bios would still need to support it, but that interface would be pretty simple.

Re:Educate me.

[Ewan]

The couple of IP based KVMs I've used are just VNC based, nothing clever about them really. They come with a rebranded VNC client to make it look like their own stuff, but you can point tightvnc, realvnc, and the others at them.

Ewan

Re:Educate me.

[ttldkns]

Thats quite confusing until you remember X11 clients and servers are the other way around (from a traditional point of view)

Re:Educate me.

[afidel]

I guess this could actually be integrated into network boot roms and add a new differentiator to the innovation stagnent NIC ASIC market.

The already have those, they are called lights out management units and they are available from every major server manufacturer. They sit in a PCI slot and provide network attached KVM style controll over the server regarless of what state it is in (even off).

Re:Educate me.

[ADRA]

That's good to know! Anyways, how does the card get powered if the system isn't on? Does it need powered ethernet or (more likely) does it use a special motherboard interface thats always powered?

Re:Educate me.

[wolrahnaes]

PCI has a low power always-on line typically used for Wake-on-LAN or Wake-on-ring systems.

That's why you may notice some NICs keep a link light on even with the system off.

I'd assume that the card uses this.

Re:Educate me.

[jeif1k]

It seems to be based on DXPC with more aggressive caching and some other improvements (e.g., it sends images in compressed form, breaks images up, and sends them at lower priority than window manager and other messages). Their mission statement says that it should work with any server and any client. It can also encapsulate RDP and RFB (VNC).

Re:Educate me.

[afidel]

They generally have a small wall wart that plugs into the card on the outside.

Re:Educate me.

[hummassa]

It gets worse because NX server is an X client and NX client is an X server and vice-versa :-P

Re:Educate me.

[Hard_Code]

Sounds like y windows.

How do things like rollovers and cursor changes work then? I wonder if this could be incorporated into X or toolkits directly via some sort of event-masking system, e.g. ALL_EYE_CANDY_EVENTS vs. YO_IM_ON_A_MODEM_SO_CHILL_OUT_WITH_THE_USELESS_EVE NTS

Re:Educate me.

[mcrbids]

I just wish there was some way to make it work at boot time so I could ditch my KVMs.

You wouldn't want to do that. I can't see how this would be more reliable than openssh - In 5 years as a full time admin, I've seen only ONE CASE where openssh wasn't working - and that's when I typed "/etc/rc.d/init.d/sshd stop"...

KVMs are there for when there's a severe software problem - the system won't boot because of filesystem corruption, a hard disk won't initialize, memory parity errors, etc.

KVM over IP raises a number of VERY SERIOUS security issues. For example, with most default installs of RedHat, you can reset the password from grub. Get control of the KVM, and you might as well have just put them inside the colo facility, given them a key to the rack, and handed them your root password. So I've not yet been able to bring myself to trust it... yet.

If I need a true KVM, I call the hosting facility and pay $0.80 per minutte. It's well worth my peace of mind... and it's rare (less than once/year) that it ever becomes an issue.

Re:Educate me.

[Bert64]

Not atall, it works like LBX... You can run nxproxy and it creates a virtual Xserver on a machine, you run your apps with the $DISPLAY set to that, and they're automatically compressed and sent over to another copy of nxproxy running on your local machine, which converts it back to standard X11 and displays the app on your local screen.. Just like running X over an ssh tunnel, except this tunnel does compression.

Re:Educate me.

[Bert64]

Or you could use decent firmware and a serial console, which everything except x86 has supported for years.. Infact i was severely disappointed to see amd64 machines using a kludgy intel style bios when they could have gone for a good firmware, or atleast a hybrid of the two like alphabios/srm.

Re:Educate me.

[Bert64]

And virtually all non x86 boxes have things like this built in, sun's have had serial consoles for years and the netra t1's among others have full LOM.

Re:Educate me.

[Bert64]

And how difficult would it be to socially engineer the hosting facility staff (many of whom aren`t very bright) into providing access into a machine..
Not only could you own someone, but you could charge them for the privilege of being owned!

Re:Educate me.

[yuri benjamin]

X11 clients and servers are the other way around (from a traditional point of view)

No they're not.
A server controls access to a resource.
A client is a program that requests access to a resource through a server.

In the case of X11, the resource is the HCI[1].
Nothing backwards about that, as long as you take it in slowly.

[1]Human Computer Interface.

Great!

[natron 2.0]

This should give them a boost in the thin client workstation office enviroment! It would be interesting to see this in action.

Re:Great!

[jav1231]

Yeah! Because that market share has really taken off! Okay, so the ever-shrinking modem-using public will finally access co-lo applications much faster.

Re:Great!

[zoloto]

You know what? That's a great idea! Instead of having KDE lag on my normal system without the network, now it can lag in user responsiveness from over the network.

either way, it's still slow

Kan we say marKeting?

[garcia]

All participants of that BoF left the meeting with a very excited feeling inside. KDE is going to really take on the struggle for the corporate desktops now, with weapons like NX/FreeNX aboard...

Are they inferring that corporations are all going to finally move to the thin-client type computing that was hyped 10 years ago? I still really doubt that it's going to happen as people are so entrenched in their current mode of deploying applications. MS Office still beats KOffice and OpenOffice and unfortunately I really don't think this is going to change that.

I read through the "interview" (which was more like a press-release-hype-sheet) and I didn't see anything that impressed me as far as non-marketing ad-speak. I haven't seen the source code (and probably won't) but I am confused as to why it must be in Bash scripts and a bunch of libraries. Why can't it just be standard code. I was especially confused by the comment "it's in shell so that everyone can contribute and make our code better." That's odd, I didn't think the Bash code did all that much if you are using libraries of machine code, etc...

I guess I will wait till someone reviews the actual code in use. Maybe after that they will rename it something less Kish than AKademy, blah.

Re:Kan we say marKeting?

[chris_mahan]

Does MS Office 97 beat OO1.1.2?

Because we're at MS Office 97 on nearly 10K desktops. (upgrade would be in the millions (which we don't have. so there we are)

Re:Kan we say marKeting?

[codepunk]

Considering I am running 400 remote x thin client terminals on the shop floor I guess the answer is yes, thin client is cool. Now of course we run bluecurve desktops and don't care much about the network efficency but I have actually tried NX and it is damn responsive over slow links.

Re:Kan we say marKeting?

[chris_mahan]

>Do you have the wherewithall to deploy and crosstrain your users?

No.

Re:Kan we say marKeting?

[stratjakt]

Yes, it does. In all respects except for cost (which you've already shouldered).

Office 97 is probably the most widely used version out there. They haven't enhanced it in subsequent releases enough to make people upgrade (not much left to enhance, really.)

Re:Kan we say marKeting?

[Otter]

Because we're at MS Office 97 on nearly 10K desktops. (upgrade would be in the millions (which we don't have. so there we are)

As a general rule, MS Office N-1 beats MS Office N.

Re:Kan we say marKeting?

[chris_mahan]

>Chances are, if you are still running Office 97, you are probably still using lower end hardware so memory consumption can be an issue.

Actually, not older hardware. All within 2 years (all win2Kpro). Just no money to update.

I mean, we have the money. Our shareholders would like us to do something more productive with it.

I hear you about OO's import. But that's also okay. We'd run both, since we've paid for 97. (btw, I love "save to PDF" in OO)

Re:Kan we say marKeting?

[hummassa]

1. no it does not.
2. we (www.almg.gov.br = Minas Gerais [3rd largest economy of Brasil] State House) just switched 700 copies of MSoffice97 for OOo1.1.2; with NO PAIN at all.

Just do it. Proper training, some care, ok, but just do it.

[]s

Re:Kan we say marKeting?

[Neil Watson]

MS Office still beats KOffice and OpenOffice and unfortunately I really don't think this is going to change that.

Please quantify that statement with facts.

Re:Kan we say marKeting?

[Cratylus]

Or you haven't seen PDFCreator, which is an open source "printer driver" that does it for Windows

Re:Kan we say marKeting?

[Eloquence]

Early this year, c't magazine, probably the most respected German computer magazine, published a quite interesting comparison of office suites. They subjected all of them to a test with very large documents with 140 pages of text, 120 graphics and 240 footnotes. MS Word apparently became less and less reliable as more pictures were added - suddenly they could no longer be moved and aligned properly without destroying existing layout. The tester gave up on Word but managed to do it with most of the other suites (including OO). They also found OO to be on the same level as MSO in terms of functionality.

Now, many bad experiences people may have with OO are probably related to importing existing MS documents. Even though the filters are pretty good, they are obviously not perfect, and last I checked macros were ignored entirely. However, that is not a fair comparison -- Microsoft would utterly fail it, as they don't have the most basic OO import filter. And the complexity of this problem is similarly high as the one of emulating the Windows API on Linux - you don't just have to get the file format right, you also have to duplicate Microsoft's way of interpreting it, even if it's buggy and/or inconsistent.

Nevertheless, the developers are always working hard on improving import filters, as it is obviously essential to business migration. OpenOffice 2.0 will have improved filters, and it will also have much better database management with support for databases directly stored in files (as Access does).

OpenOffice is clearly more performance-hungry than MS Office, although in my experience that is mostly the start-up time. I don't anticipate major improvements in this area. If you're looking for a very slim MS Word replacement, KWord or AbiWord are probably projects worth keeping an eye on. TextMaker, a proprietary package, also exists for Linux. And if you're into DTP, Scribus is quite mature already.

Re:Kan we say marKeting?

[bluekanoodle]

Can we give the man a pat on the back for his honesty?

Re:Kan we say marKeting?

[AvitarX]

"They haven't enhanced it in subsequent releases enough to make people upgrade (not much left to enhance, really.)"

By enhanced it enough do you mean changed the file format?

Re:Kan we say marKeting?

[Maxwell]

>Since you have been sleeping under a rock for the >last 10 years, you should probably make a search >on the net for Citrix.

>>There are nothing but complaints about Citrix.

Thats your problem, not Citrix. Citrix is becoming increasingly popular in large corp environments, particularly healthcare. Many, many hospitals are 99% citrix based thin client on the floors.

If you haven't worked at at least 20 different sites that have over 1000 PC's each, you haven't worked in the target citrix market. In that case, you are not in any way qualified to comment, are you?

JON

Re:Kan we say marKeting?

[Maxwell]

>Do you have the wherewithall to deploy and >crosstrain your users?

I am not sure the cross training is a factor.

I work in a perpetually cash strapped industry.

One of my clients did a 'study' (not scientific, but they spent a few weeks on it, with 12 people) on the migration costs from Office 97. Half the gorup migrated to Office XP and the other half migrated to OO1.0 (pretty sure it was 1.0, this was in early 2003)

They found it was easier for the end user to migrate to OO then to Office XP. Lots of confusing stuff in XP and many changes just for the sake of changing. They also scored exactly the same on the document migration process - yes XP wouldn't always get an O97 word doc right!( The spreadsheets worked 100% in XP, not very well in OO.)

They went to XP, MSFT gave them a ~$21/seat deal because they are a non-profit. Part of the deal was they had to buy a license for every seat, whether it needed Office or not. Sound familiar?

It's not a training issue as 97 and XP are APPARENTLY just as different as 97 and OO

JON

P.S. small study. meaningless. I know. But think about it...

Re:Kan we say marKeting?

[chris_mahan]

I will, thanks.

Re:Kan we say marKeting?

[Enahs]

I hate to pick nits, but OpenOffice already supports databases directly stored in files. It's just that they're dBase files.

Re:Kan we say marKeting?

[yuri benjamin]

even smaller study - my own household:

Wife: Win98 + O97 -> KDE + OOo
Dad: DOS + MS-Works -> KDE + OOo

Wife - no problem. Dad - still adjusting to newfangled gui, but getting there. Dad's proof that guis are not for everyone (coming from MS-Works for DOS, which still works in DOSemu).

So...

[stratjakt]

This is PCAnywhere for linux?

I run sessions over shit dial-up connections, like 16.8kbit or so, and the responsiveness is decent. If I get a full 56.6 connection, it's really good.

I know we cheerlead for OSS around here, but is this a brand new amazing wonderful thing, or just another VNC protocol? And does KDE need more stuff? The K is for "Kram it all in!"

Re:So...

[Aliencow]

PC Anywhere is a slow piece of shit, don't tell me you get decent responsiveness with decent resolution on a 56.6k line.

Or maybe you have different standards than anyone.
RDP is decent, PC Anywhere = piece of shit.

Re:So...

[stratjakt]

Maybe you haven't used it recently (version 10+), or havent paid for the commercial licensed version, and have the crippled crap edition, or you don't know how to tweak it?

With the type of stuff I use it for, 4 color mode is fine, and it works well. I've found PCAnywhere's file transfer mode is faster than ftp too.

Re:So...

[Doctor Crumb]

PCAnywhere for linux is called X11. It's been around for years and years, this is just KDE taking advantage of a protocol to speed that up.

Re:So...

[Com2Kid]

• RDP is decent

Decent my ass, over a 1MB down 256Kbit up cable modem connection, RDP blows massive chunks.

Type

Wait wait wait wait

Type

Wait wait wait wait

Click

Wait await wait wait.

Horrid.

Heavily JPEG compressed VNC isn't much better.

RDP over a 10mbit LAN? Still a bit jerky, but at least usable.

Re:So...

[TomorrowPlusX]

There needs to be a "No concept of history" moderation.

X11 has been around for longer than many slashdotters have been alive.

And does KDE need more stuff? The K is for "Kram it all in!"

Oooh! Another joke about KDE having too much functionality, and a clever wordplay on the letter K!!! Take that KDE developers!!!

Re:So...

[danheskett]

What are you doing wrong? I am connected to my home workstation quite a bit, and my router uses QOS to limit RDP packets to no more than 8 kb/s. I use 1024x768x16 without any problem. It's fast, wicked responsive (I forget where I am sometimes), and the latency is great.

Granted, I have a nice connection back to home - my average ping is about 15 ms.

I'd bet you are seeing bad response due to connection latency.

Re:So...

[XMyth]

I can't attest to RDP over a dialup as I've never used it, but I have used it over a DSL connection(IOW 10mbit lan) and it was just perfectly fine. No problems whatsoever. Something was wrong with your setup.

Re:So...

[misleb]

8kb/s??? My guess is that your QoS isn't really doing anything.

Re:So...

[Com2Kid]

• I'd bet you are seeing bad response due to connection latency.

10ms latency.

Or less.

My friend and I are both on cable modems, we live blocks away from each other.

800x600@8bit is dirt slow. 800x600@8bit over a 10mbit connection is STILL jerky.

This is using Netmeeting mind you. Getting Terminal Services working proved to be, umm, futile. (3 OSs latter. . . . we curse MS's licensing scheme!)

Re:So...

[Com2Kid]

Well I don't imagine it is too dependent upon CPU speed, the system being connected to is a 600mhz Celeron with 384MB of RAM.

As I said, netmeeting, dialing into. Everybody is behind a NAT, but I have HUUUGE holes punched in it, ports forwarded to the appropriate client and server machines.

I have yet to see remote desktop work very well on Windows XP, what system are you using? I know that Terminal Services kicks Remote Desktop's arse, but MS won't let us at it. >:O

Re:So...

[XMyth]

Actually I've found Terminal Services to be about the same to Remote Desktop (infact I'm pretty sure they're identical...terminal services client will connect to a remote desktop).

I've used Remote Desktop across dsl to a 1.5ghz P4/256megs ram.

I've used Terminal Services across cable to a dual 1 or 2ghz server with probably a gig of ram....can't remember specific machine specs. Both worked great.

I tell you what works even better though, Gotomypc. That is a really nice app there. Reverse connections too.

VNC really isn't that bad either if you've got the mirage driver installed (otherwise the server eats up CPU like crazy). You can even set that up to reverse connect too. I've written an app to help automate that actually....if you're interested in that app let me know...I believe I have the (C#/.NET and some perl) source online at rootedbox.com somewhere.

wow distributed computing

[bkruiser]

who would have thought!?!? No need to send Aunt Martha to personal computer administrator classes at the local community center!

Wait isn't open source supposed to only copy

[number6x]

Wait isn't open source supposed to only copy already existing closed source technology?

How dare they be innovative.

Next thing you know linux will have a measurable and growing market share.

Then software like apache, eclipse, and jboss will be used in enterprise applications.

Oh, wait...

Never mind!

Re:Wait isn't open source supposed to only copy

[number6x]

The post was really a feeble attempt at humor, but since you bring it up...

What I've read about NX is that it is more like x11 than it is like a VPN. Linux does have several VPN implementations to choose from, but NX is really a lightweight windowing client / server protocol.

Microsoft might have tried something like this back in the Win 3.11 days, but all they did there was make some of the office products client / server capable. After the shift to NT based os's, and their move to a peer to peer network, I think that they pretty much closed the door on something like X11 or NX (that's not a criticism, going client server has as many issues as going peer to peer, just different ones).

Besides, as Microsoft has shown us innovation is for losers. Winners let the Apples and Suns of the world innovate. Then the winners merely implement the stuff when it becomes a common standard. That way you don't waste time and money on R&D.

Re:Wait isn't open source supposed to only copy

[Gooba42]

VPN or VNC?

Re:Wait isn't open source supposed to only copy

[Shinobi]

More to the point than you might think.

NX actually uses RDP to achieve the low-bandwidth solution that is still capable of displaying a whole desktop and still retaining responsiveness.
And Microsoft put a huge amount of research into RDP, and it's what they've been using for their Terminal Server and Remote Desktop applications for a _long_ time now.

My personal experience is that Remote Desktop from my home box is more responsive displaying the entire desktop and all apps(fuldc++, Ultraedit, Miranda, Virc, CodeWarrior) than X11 is with just one app(Codewarrior for Linux) from the same machine to the same machine via the same connection.

Re:Wait isn't open source supposed to only copy

[suckmysav]

"What I've read about NX is that it is more like x11 than it is like a VPN. Linux does have several VPN implementations to choose from, but NX is really a lightweight windowing client / server protocol."

I have used NX and you are both sort of correct. Basically, NX is a compressed X11 session that communicates via SSH, which I guess is what the VPN stuff was a clumsy reference to.

It is a neat feature, because all you need to do to remotely control a host on the otherside of a firewall is to open a pinhole on port 22, and you can connect via either SSH on the CLI or as an graphical NX Client.

It is a pretty good solution really.

Re:Wait isn't open source supposed to only copy

[Nurf]

More to the point than you might think.

NX actually uses RDP to achieve the low-bandwidth solution that is still capable of displaying a whole desktop and still retaining responsiveness.
And Microsoft put a huge amount of research into RDP, and it's what they've been using for their Terminal Server and Remote Desktop applications for a _long_ time now.

My personal experience is that Remote Desktop from my home box is more responsive displaying the entire desktop and all apps(fuldc++, Ultraedit, Miranda, Virc, CodeWarrior) than X11 is with just one app(Codewarrior for Linux) from the same machine to the same machine via the same connection.

I bought NX a while back and use it to control some of my machines when I went to South Africa for christmas. I used it to run an Gnome - Evolution session on a machine here in Seattle.

NX does not use RDP for X sessions. NX uses careful caching of the X11 protocol, along with a library that pretends to be xlib. NX is capable of handling RDP, but then it uses its own optimised version of RDP. It doesn't use RDP for X. I'm pretty sure about this, because I downloaded the GPL source they provide and made a working version (which was not easy. :-) ). I eventually bought it for the convenience of the wrapper applications they provided.

X11 has serious issues with slow links, not because of bandwidth requirements, but because xlib requires a stupid number of synchronous queries for many operations. One estimate I have seen, is that around 300 X packets have to make a round trip before a window will open (in older versions of X, without the Render extension). On a local session, this is all short circuited and you don't see any delay. On a link with (say) a 300ms delay, you are looking a 300*300ms delay, even though each packet is minute.

Keith Packard has done a lot of work on adding extensions to X that mitigate this. Also, there is work on writing a library the implements the X protocol that is completely asynchronous and caches data when possible (which is sort of what the NX .so replacement for xlib does).

NCB can be found here, at http://freedesktop.org/Software/xcb.

I agree with you. Current X is dog slow over high latency links.

Soon, we'll have the XDamage extension too, which will help a lot. It's at http://freedesktop.org/Software/XDamage

As an aside, I am very pleased we have moved away from XFree. X.org should be a massive improvement.

More like a revolution in

[GillBates0]

Network Komputing

/ob joce

Re:More like a revolution in

[Soul-Burn666]

Don't you mean kNetwork Komputing? (the k is silent)

How possible

[elgatozorbas]

How can you transfer a browser's contents over a 40kB line when its own internet connection can be a lot higher?
Z

Re:How possible

[Richard_at_work]

Post browser rendered data might be faster to transfer. You dont have to transfer what the browser actually downloads.

Re:How possible

[IvoryRing]

I'm going to give you the benefit of the doubt and assume you can't really see any way this can work:

• Image representation that is more space efficient than the source. For example, someone uses 2000x2000 GIF file but sets the size on the IMG to be 200x200 (There are people that do this to 'precache' in case you want to look at the full sizes - no it isn't very helpful, but it happens) - resampling that down before sending it will send a lot less data. Also, on a page full of JPGs, if I am willing to waste CPU cycles, I can always recompress them to lower quality. I may, as the viewer, be willing to make the quality tradeoff, but others on a full speed connection don't have to suffer with it (as they would if the page author down-sized)
• Bloated HTML. There can be tons of junk - either intentionally (think spammers attempting to get around word matching) or accidentally (certain HTML writers are really much more verbose than they need to be - for example, explicitly listing out every option to a tag even when the defaults are what you want; also it isn't unknown to see two or three different blocks of scripting to handle different browsers)
• Significantly long pages. If a page contains 30 screenfulls (yes, poor design, but that doesn't mean it doesn't happen - remember that you have no control over the source) but everything I need is on the first two (or first and last), there is no real reason to send all 30 pages to me.
• Other reasons where the source material is significantly more data than the resulting rendered text

Obviously this doesn't always work, but in many cases it can work very well.

Been there, done that

[xcomputer_man]

Evas-based apps (including the upcoming E17 window manager) perform extremely well over remote X connections, using traditional Xlib. I have tested this myself, over remote connections Evas-based apps are at least 10 times more responsive than GTK/QT apps, using the same traditional X11 connection. Evas is designed to minimize roundtrips to the server so everything gets drawn the first time. And there's a new canvas server in CVS called Evoak that allows remote canvas sharing among applications, complete with gzip compression etc....NX probably won't even be able to touch it performance wise.

Re:Been there, done that

[dago]

Does that works for any random X application without modification ?

Re:Been there, done that

[On Lawn]

I have to admit I've seen xeyes and Koules perform much better than even the most simple GTK apps over remote connections. It seems athena and other basic toolkits do better than these newfangled ones ;)

The GTK apps also seem to benefit the least from X compression schemes I've tried. Well the ones that aren't screen-shot deltas like VNC and NX. Actually I'm only partly sure that NX uses delta compression but it performed great.

I'm glad to hear that E is designed from the ground up with remote connections in mind. I still remember the visually stunning themes back in the day. (My favorite being the default plastic looking one that looked like a light was shining on applications and buttons that were active). Though I like GTK, I've found remote connections to be its greatest weakness in my daily routine. I never did try E over remote connection back then.

By the way, since I'm busy and lazy (go figure) does anyone know if FreeNX does not require a seperate login database like NX does?

Re:Been there, done that

[xcomputer_man]

Does that works for any random X application without modification ?

Apparently you weren't even reading. I said "Evas-based apps".

And while NX is still a hack that attempts to get around the problem of poor network performance for X11 apps (by wrapping/translating X calls with its own calls), Evas solves the problem directly by making efficient use of Xlib and eliminating unneeded roundtrips. Evoak, the shared canvas goes even further: the host canvas is running on the same machine as the X server, and clients connect to Evoak (which has a much leaner protocol, supports compression *and* encryption) rather than to X. This is the kind of technology that could make running remote X apps on your cellphone possible if resources were available to implement it.

If you want to find out more go read http://enlightenment.org/pages/systems.html.

Re:Been there, done that

[dago]

Not every question asks for an answer.

Re:Been there, done that

[Bert64]

Ofcourse theres no way to beat well designed toolkits, unfortunately most of the common toolkits are NOT well designed and cause massive amounts of network load.. This may be fine on localhost or over a 100mb switched lan, but over a wan link it's terrible..
I used to use X11 apps over dialup (very slowly) and a 2mb line (very responsive) a few years ago, but modern apps over another 2mb line with slightly less latency are unuseable.

thin clients revisited

[cjsteele]

My employer had previously deployed 2,000 modified NetBSD thin clients from IBM that ran off of 200+ Linux boxes that provided the OS, print and storage facilities, but let the thin client do the grinding on the apps... only difference here is that the thin client doesn't grind on the data, just renders screen shots. Fact of the mater is, both approaches are highly manageable ways to provide low-cost computing to the masses...

Re:thin clients revisited

[gregmac]

My employer had previously deployed 2,000 modified NetBSD thin clients from IBM that ran off of 200+ Linux boxes that provided the OS, print and storage facilities, but let the thin client do the grinding on the apps... only difference here is that the thin client doesn't grind on the data, just renders screen shots. Fact of the mater is, both approaches are highly manageable ways to provide low-cost computing to the masses...

Sorry to be OT, but do you have links to any more information on this? I've always liked the idea of thin clients from an administration perspective, but they're always such a waste of resources. If you have a bunch of desktops that are all, say, 1ghz machines and turn them into thin clients, it's a pretty big waste of power -- not to mention money, since you're now duplicating that power on an expensive server.

It would be nice to have the systems interchangable (one breaks, swap in another) like thin clients, but at the same time, use the processing power of the thin client to allow the server(s) to be less powerful (cheaper), or allow fewer servers to do the same job.

Re:thin clients revisited

[tmasssey]

Workspace on Demand.

IBM had, what was to me, the best thin-client system on the planet. It was heavily OS/2-based, but there was a version for Windows as well! (First link above). Full fat clients running full fat operating systems, but served, managed and administered from a central server. Hardware dies? Use a different computer: get the same operating system, environment, etc. Roaming user? Get *your* OS from any computer. Need to deploy an app to 1000 users? Install once, and drag and drop an icon. Done.

To me, *that* is Network Computing. Not glorified VNC to a single computer...

Re:Konquerer, Mozilla, and KMail...

[RevAaron]

Mozilla doesn't have a file manager that integrates well into the desktop env as a whole.

Re:Not Any Time Soon

[XMyth]

Are you sure this can't be implemented at the toolkit level? That would certainly make adoption easier.

Re:Not Any Time Soon

[RyuuzakiTetsuya]

Just have one of the heads of the KDE project get on a stage at a Linux expo sweating and chanting, "Developers!"

Re:Not Any Time Soon

[dago]

Warning : Acronym Collision

The correct moderation to apply to the parent post is either "Offtopic" or "Funny", the latter being more my choice

Quick karma whoring :
- AMD NX : No Execute, prevention of buffer overflow (stupid webpage here, search google for AMD NX)
- nomachine NX technology (website), which is, functionnality-wise, the sucessor of VNC

Wow! Innovation!

[Phat_Tony]

Soon, maybe they'll invent "dumb terminals" that run all their programs off a central, "mainframe" computer!

Re:Wow! Innovation!

[YU Nicks NE Way]

Wow! That's a cool idea! And then, we could add a graphics protocol and send commands for that over a narrow pipe to the terminal, like Windows' RDP...I mean, like NX, that revolutionary newly thought up idea does!

Re:Wow! Innovation!

[xutopia]

Revolutionary Rev`o*lu"tion*a*ry, a. Cf. F.
r'evolutionnaire.
Of or pertaining to a revolution; repackaged product.

Re:Wow! Innovation!

[Detritus]

Mainframe terminals are not "dumb". They allow the user to fill out and edit the contents of a form without using any CPU cycles on the mainframe. The mainframe receives a completed form, not hundreds of individual characters.

Re:Wow! Innovation!

[sketerpot]

NX isn't a new idea, it's a good implementation. They claim to be bringing X into modern times. And they support RDP, as well as whatever VNC uses.

Re:Wow! Innovation!

[aminorex]

Hey, what goes around, comes around.

Re:Wow! Innovation!

[Bert64]

Actually, it's a newer and better version of LBX, which predates windows RDP by many years... This is just an improved version of something that has long existed.. its windows which copies from others and never has anything innovative (aside from email viruses)

I don't believe

[Snaapy]

into a 40 KBit/sec modem connection without losing responsiveness for the user experience.

I'd like to see any responsive image data over 40 Kbit line. Let's imagine some icon takes 64x64 pixels and is crunched to 4KB compressed. It still takes 0,5 seconds per an icon to load. Opening a start menu, waiting... please be patient.

Anyone who have surfed on a modem knows it's far from real-time responsiveness.

Re:I don't believe

[omega9]

Let's imagine you had a clue what you talking about...

You could've made it easier on yourself and just said "I've never used this stuff so I have NFC what I'm talking about."

Re:I don't believe

[the chao goes mu]

Given your numbers, 0.5 is wrong.

4KB=8Kb*8=32Kb.

It would take 32/40 or 0.8 s

Re:I don't believe

[the chao goes mu]

And my own correction is wrong.

I meant to say 4KB * 8, not 8KB * 8.

The 32Kb is correct though, as is the 0.8s.

Re:I don't believe

[misleb]

Well, the way X (and I assume NX) works, many of the images an application needs are copied to the client (server, in X11-speak) to be stored in RAM during startup. Then they are pulled out of memory to display as needed. The image is only transfered once unlike with VNC, PCAnywhere, etc. What you end up with is a longer startup time while keeping the UI responsive. Of course, this depends on when the application decides to load pixmaps into RAM, but generally it works pretty well.

Also, a 64x64 icon will usually compress to WELL BELOW 4k. Transfering it would be trivial. And as others have pointed out, it wouldn't take 0,5 seconds to transfer 4k over a 40k/sec link. Try 0,1 seconds. ;-)

-matthew

Re:I don't believe

[misleb]

Oops, forget that silly calculation. I was thinking bytes and not bits.

Re:I don't believe

[misleb]

What you don't seem to understand is that this technology doesn't transfer the pixmap when it is displayed on the screen like VNC or PCAnywhere. THe pixmap is stored in the RAM of the local machine and an instruction to display it is what is passed over the modem link. This amounts to a longer application startup time, but the responsiveness remains reasonable.

-matthew

Re: I don't believe

[Omniscient Ferret]

Let's imagine some icon takes 64x64 pixels and is crunched to 4KB compressed.

This would be a "compression" of a byte per pixel: 64x64=4096=4K. Compressing with JPEG or PNG would make it a fraction of that size.

Re:I don't believe

[iantri]

Your math (as other posters have mentioned) is wrong; but never the less it does buffer icons and that sort of thing locally.

The second time you open the menu it is far faster.

Edumacation right here.

[UnderScan]

Introduction to NX technology
A brief introduction to NX motivation and technology
This document outlines the background and the design decisions that guided NX development. It explains why NX is different from similar technologies and states the goals the NX project is set to pursue.

Re:How's that different from how things work now??

[codepunk]

Have you actually tried a connection using NX? I have and I can tell you that is damn responsive remote x even over a slow link. I run 400 remote x terminals on a lan and it is not much use for me but if I needed to support road warriors I would not hesitate to use it.

Yay, another overloaded acronym....

[Xentax]

I know we're running out of possible acronyms that don't already stand for something, but releasing two new 'overloads' for an acronym almost at the same time sucks.

(If you're wondering, we have this NX client software, and the NX 'No-eXecute' flag on CPU's to help contain the threat posted by stack and heap overflow vulnerabilities)

We're running out of TLA space a lot faster than IPv4 space. Not as big a deal, I know, but just wait until companies start trying to brand/trademark acronyms or initialisms (for the purists out there) when there's already existing meanings for their choices...

Xentax

Re:Yay, another overloaded acronym....

[currivan]

What a great name for software. Does this remind anyone else of the "No Va" urban legend?

Re:Yay, another overloaded acronym....

[Xentax]

The scary thing for me was seeing the "No Va" urban legend in a *college textbook* on 'Organizational Psychology' I think it was? You'd think I'd remember the last college class I took.

Oh well. I guess including an acronym or initialism in your brand is just part of modern 'buzzword compliance'...

Xentax

Re:Yay, another overloaded acronym....

[TeknoHog]

NX is not an acronym, it's an abbreviation.

Acronyms are a special kind of abbreviation that you can pronounce as a word: for example BASIC, NATO, LASER.

Re:Yay, another overloaded acronym....

[Skater]

I've been advocating a switch to SLA for years now. Like IPv6, it will give a LOT more room for acronyms. Six Letter Acronyms today! Join the cause.

--RJ

Re:Yay, another overloaded acronym....

[alerante]

"TLA" is overloaded too. Do you mean two- or three-letter abbreviation? ;)

Re:Yay, another overloaded acronym....

[jeorgen]

I know we're running out of possible acronyms that don't already stand for something, but releasing two new 'overloads' for an acronym almost at the same time sucks.

(If you're wondering, we have this NX client software, and the NX 'No-eXecute' flag on CPU's to help contain the threat posted by stack and heap overflow vulnerabilities)

We're running out of TLA space a lot faster than IPv4 space.

You realize your use for TLA as Two Letter Acronym overloads TLA meaning Three Letter Acronym?
:-)

Re:Yay, another overloaded acronym....

[Xentax]

This has come up in debates several times where I work.

I have yet to find what I'd consider an *authoritative* statement on exactly what constitutes an Abbreviation vs. an Acronym, vs. an "initialism".

First, AFAICT, an abbreviation is always a shortening of a single word. Please correct with a counter-example if I'm being a 'tard there.

Second, to *most* people, again AFAICT, an acronym is a mostly-first-letters-shortening of a phrase or group of words.

Whether this is limited to "pronounce-able" shortenings (NATO, LASER, RADAR, MODEM) is pretty debatable, though I've seen "initialism" defined as a term describing those that *can't* be pronounced (USA, XML), thus either separating or at least distinguishing them from those that can.

Of course, the waters are muddy even outside of that. For example, must they only be first letters? RADAR (RAdio Detection And Ranging), MODEM (MODulator/DEModulator) would suggest otherwise. Need every (non-trivial) word be included? "Interpol" (International Criminal Police Organization) has been listed as a sample acronym online, so maybe not. Even "start of word" isn't *always* apparently required - consider DNA (DeoxyriboNucleic Acid) and eXtensible Markup Language. These seem to be handy exceptions to a general rule, but they're still pretty popular.

Finally, there are the ones that, depending on who you ask, are either pronounce-able or not. SQL (is it "S-Q-L" or "sequel"?) and URL (U-R-L or "earl") are two examples of this.

Like I said, it's something I've discussed/debated/Google/Wiki-Pedia'd over before.

Xentax

Re:Yay, another overloaded acronym....

[Xentax]

That's certainly one way an author can try to avoid the confusion. It was a "strongly suggested" practice when I was in the academic habit of writing essays.

In general, though, it seems like you operate in a ... subculture, environment, whatever ... where, HOPEFULLY, only one meaning at a time would be likely (especially given any sort of meaningful context).

Xentax

Re:Yay, another overloaded acronym....

[Xentax]

Yeah. Which is funny considering I think of three-letter-acronyms when I write/say "TLA", despite the fact that my post was about NX, a two-letter-acronym.

With apologies to CCR (hah!), who will stop the [madness]?

Xentax

Re:Yay, another overloaded acronym....

[TeknoHog]

The -nym suffix means 'word', as in 'synonym'. I think a word is something that must be pronounceable.

Wikipedia tells that it makes no difference if the abbreviation in question is an initialism or not. What matters is the pronunciation.

Re:Yay, another overloaded acronym....

[rvega]

An acronym does not have to be pronouncable, it only has to be formed from the initial letters of the original words. The examples you all give are, indeed, acronyms. But so are RSVP, NRA, SMB and WTF.

Re:Yay, another overloaded acronym....

[Suidae]

It might be entertaining to write a small script to find all the reoccuring three and four letter word combinations in a document and automatically search-n-replace them with an appropriate acronym, using the parenthetic notation for the first replacement.

This could have some entertainment value when used on long term papers.

Re:Yay, another overloaded acronym....

[MrCreosote]

'SQL (is it "S-Q-L" or "sequel"?)'

Squeal.
(Cue 'Deliverance' banjo music)

Re:Yay, another overloaded acronym....

[aminorex]

> I think a word is something that must be pronounceable.

Like 0x77CF?

Anything you pronounce is demonstrated thereby to
be pronouncable. I admit that XML isn't pronounced
kzmul, but rather eksemel, but it's still being
pronounced. The word knight isn't pronounced
kunigit (as it was in the 1300s) either. That
doesn't mean it's no longer pronouncable. It
just means the pronounciation has diverged from
the standards more commonly applied to similar
sequences of letters.

Re:Yay, another overloaded acronym....

[aminorex]

If you are going to take the Mikkelson's whiggish
ideological rants on their sensationalist website
as a higher authority than your
board-reviewed psychology textbook...well...

I guess it takes all kinds to make a world...
suck.

Re:Yay, another overloaded acronym....

[drfreak]

However, that is also an overloaded acronym. How much namespace does a Single Letter Acronym get you? Depends on which alphabet you use I suppose..

Re:Yay, another overloaded acronym....

[Xentax]

Actually, even my "language requirement" level of Spanish comprehension was sufficient to show the fallacy behind the "no va" urban legend.

And I find the idea of elevating that book to "board-reviewed" and thus implying some sort of infallibility to be downright *hilarious*.

Anyone who doesn't realize that nearly everything anyone ever "teaches" you is WRONG and/or not so cut-and-dry, has a lot of growing up left to do.

Xentax

Re:Yay, another overloaded acronym....

[Xentax]

An acronym does not have to be pronouncable

Uh...because you said so, or what? Did you even read my post?

It started with something to the effect of "I haven't been able to find a truly authoritative source" as to whether an acronym MUST be prounceable or not. And I *have* looked.

If you're going to take a stand on the subject, let's at least hear about why you think so - cite a reason, a source you consider authoritative, something.

Xentax

Re:Yay, another overloaded acronym....

[rvega]

Hmmm, that's strange. I only read your post just now. I thought I was replying to the same parent you replied to, in which context my post makes better sense.

In any case, Merriam-Webster's definition indicates that an acronym may be either pronouncable or not. Personally, I would personaly consider Merriam-Webster to be the second most authoritative English dictionary, after the OED, which even bills itself (probably rightly) as the "definitive record of the English language". Unfortunately, OED doesn't allow free online lookups and, again unfortunately but not surprisingly, "acronym" does not happen to be their free word of the day.

Other online dictionaries disagree about pronouncability as a requirement. I like what another poster said about the "nym" suffix meaning "word", but I don't know if I agree that a "word" must be pronouncable. I haven't really thought about it. Lots of words don't agree 100% with their etymology (the chairman might be female, for example.)

I agree with you that an abbreviation should be the shortening of a single word, but some dicionaries seem to indicate that an acronym is a special form of an abbreviation, which disagrees with our point of view.

If I could get OED's verdict I would submit to whatever it happens to be. Failing that, I'll cite Merriam-Webster, which happens to accord with my own opinion.

Re:Yay, another overloaded acronym....

[Xentax]

FWIW (hehe), I tend to agree with you - both that Merriam-Webster is at least reasonably authoritative, and (to put it in my own terms), the common *understanding* these days seems to include both ones you can pronounce as well as those you can't (NATO and FBI, to choose random examples of each respectively).

I like "initialism" as a *subset* of acronyms - those that can't be pronounced - rather than a disjoint set. But that's strictly my personal feeling.

The whole word thing sort of came out of left-field to me. But I have to agree to the extent that a term's etymology may have very little to do with it's modern, common usage and meaning (see geek and virus among other terms whose meanings have changed substantially over time).

Of course, people get bent out of shape about how common usage corrupts the 'true' meaning ("irony" being the current flame-inducing term of choice). I try not to encourage that kind of thing in my personal communication, but I'm not gonna be one of those who fights to the last ditch to save "The Queen's English", either.

Xentax
ps. I have been known to tee off on people who say "irregardless" though ;)

Microsoft had this for years :-)

[melted]

You can do RDP over analog modem and things are pretty darn snappy. And you could do this since Windows NT4 Terminal Server edition. Remote desktop comes stock with NT OSs since Windows 2000.

Re:Microsoft had this for years :-)

[omega9]

To clarify the parent a little,

The client side comes stock on NT, 2k Pro and Server*, XP Pro and 2k3 Server*. However the MS RDP client is downloadable for free from their site.

The server side only comes stock on NT,2k, and 2k3 Servers, not the workstation OSs. And even then, you have get a single "stock" license, so no more than one connection at a time unless you shell out some bucks.

Re:Microsoft had this for years :-)

[DeDmeTe]

Ya.. Citrix invented it!

Re:Microsoft had this for years :-)

[slittle]

Um... remote desktop client and server is definately in XP Pro (along with remote assistance). Of course it's only good for one user at a time (I assume this is an artificial limitation). What comes with the Servers is likely the full dealie that allows multiple concurrent desktops.

Re:Microsoft had this for years :-)

[HawkingMattress]

To clarify more :

XP pro has TS server, you just have to activate it in "my computer". But only one person at a time can be connected (including the console, so if someone is sitting in front of the computer he will be suspended). Still, it's quite handy...

And i don't know about others, but on 2k server with the included licence you can have 2 connections at the same time, not one. Plus another one at the console.

Re:Microsoft had this for years :-)

[HawkingMattress]

It's the fastest thing i tried, and i've spent days working via remote graphical connections on a slow line. (client DSL at home -> server ISDN )

To sum up my personnal benchmark :

• VNC... a dog
• remote X... highly depends on the way the app is coded, seems faster than VNC in some instances, 100x slower in others (just checked this one for fun, my work was mostly with windows apps so i'm not too sure...)
• PCAnywhere.. about 2 times faster than VNC
• Citrix, Terminal Server: easily 10x faster than pcanywhere, probably more. Most of the time you barely notice you're on a remote machine.

Re:Microsoft had this for years :-)

[dsanfte]

You guys realize there's no license verification in 2k server, right? You can open up licensing and increase that number of licenses to your heart's content -- hell 100,000 if you want -- and there's no actual "checking" to see if you have them, just a checkbox and some legalese saying "if you don't have these licenses, you're in trouble if we audit you".

Re:Microsoft had this for years :-)

[HawkingMattress]

Yep, but we're talking about the number of users you can legally use with the bundled "standard" licence.

Re:Microsoft had this for years :-)

[ByteSlicer]

I assume this is an artificial limitation

Indeed, the winlogon process does an RPC call to the termsrv module to check if an additional connection is allowed. Termsrv is statically linked to the licensing module. Some beta versions op SP2 enabled multiple connections. More specifically, replace \WINDOWS\SYSTEM32\TERMSRV.DLL with version 2.1.2600.2055 (find on internet, replace in command line safe mode) and the registry key below to enable it (Requires Windows XP SP2 (RC), fast user switching, non-empty user passwords).

[HKEY_LOCAL_MACHINE\ SYSTEM\ CurrentControlSet\ Control\ Terminal Server\ Licensing Core]
"EnableConcurrentSessions"=dword:00000001

Re:Microsoft had this for years :-)

[onpaws]

Also so does Windows XP Professional with its remote desktop feature... Even my favorite OS, OS X, doesn't have anything nearly as fast as this (although OS X is a much more bit-mapped OS)

Re:Not Any Time Soon

[Anonymous Coward]

NX will be useless until a sufficient portion of programs support it.

You are completely right in that, mate!

However, please take note of the fact that virtually all (meaning 100%) of all X11 programs already support it. I have used NX successfully with KDE (each single damn program of the lot), GNOME (most of their programs -- I have abandoned GNOME as my default desktop a year ago), ICEwm, OpenOffice.org, Acrobat Reader, Mozilla, Firefox, Abiword, and a bunch of others. They all worked.

The reason is simple: NX uses the X11 protocoll (which each X11 program uses) and translates it into its own NX calls to bridge the remote link distance. After the bridging, it re-translates it into X11 and, voila!, the local X-server displays the X app's GUI without a hitch...

Teacher (aka non-commercial Tutor) here.

[SlashdotTroll]

This is using KDE's widgets; the article does not conclude with any reasonable technical confrontation of the X11 protocol. The feature is KDE can be used to minimize the Xlib transport layer by using only widgets. It is a verry impressive feat, the general purpose of X11, but this is using KDE libraries which are slowly demanding more system resource overhead just to run. The largest gripe I have with KDE is it is more difficult to jump between KDE widget context and Xlib context. I've been able to program using Xlib and Xtk for the past 4.5 years and all the bloat in KDE is justified, but I fear X protocol will slowly be over-ruled to accept legislated features previously extensions. X is meant to be a verry quick drawing canvas for low-bandwidth connections. Next thing you and I will know is they will be calling it XKDE or XGNOME.

Re:Teacher (aka non-commercial Tutor) here.

[AvitarX]

What about Mozilla and Open Office?

Re:Teacher (aka non-commercial Tutor) here.

[misleb]

X was doing it years before Microsoft in a much more modular and flexible way. KDE/NX is improving on that, not imitating Microsoft. Terminal services require you to pull a full desktop from a remote server. X allows one to pull individual single applications that display as if it were running local. WAAAAYYYY better than Terminal Server IMNSO.

Re:Teacher (aka non-commercial Tutor) here.

[eean]

I'm pretty sure the parent is BS or I just can't read what its saying. NX takes the X protocol and uses various caching and compression methods to make it more efficient. Unlike VNC which essentially takes a picture of you desktop and sends that, so its easy to see why NX works so much better. It is desktop environment anogistic more or less The client and whatnot I think are written using KDE widgets for the config menu, it uses artsd for sound. The folks developing the Free version are KDE people, thus its under the KDE category. But the server certainly doesn't know anything about KDE or Gnome or whatever, it just deals with X.

X when using xeyes, xconsole and twwm might be a quick bandwidth-efficient drawing canvas... but it isn't with any modern program, thus the need for something like NoMachine.

Why do you want to mix KDE and Xlib? Folks developing KDE don't even use Xlib, they leave that up to Trolltech. The only program I can think of thats still developed (sure there are others) and uses Xlib is the mplayer GUI, and I think everyone accepts its a POS, it mixes xlib and GTK last I heard. I just use mplayer from the console, sometimes with one of its KDE frontends.

Re:Teacher (aka non-commercial Tutor) here.

[Rich]

I use xlib directly in a few of my apps like KSnapshot. Getting hold of an X11 window or paint handle is a one liner. 'Slashdot Troll' is talking out of his ass.

Re:Teacher (aka non-commercial Tutor) here.

[greenrd]

I'm pretty sure the parent is BS

Well done for spotting that SlashdotTroll is, uh, a troll!

Re:Teacher (aka non-commercial Tutor) here.

[blackpaw]

Terminal services require you to pull a full desktop from a remote server.

Wrong
X allows one to pull individual single applications that display as if it were running local

So does terminal server
WAAAAYYYY better than Terminal Server IMNSO.

Try X over a modem or even adsl connection - its painful whereas TS and Citrix are very usable

Re:Teacher (aka non-commercial Tutor) here.

[omicronish]

On the other hand, I do like being able to "save" my current work state on the server and reconnect later and continue with my programs still open. I'm sure you could hack something similar with X, but terminal services isn't as bad as you make it sound.

Re:Teacher (aka non-commercial Tutor) here.

[eean]

I wasn't aware KDE programs did that, though it makes sense for KSnapshot. Seems like it will make it tough for the KDE windows porting effort... though I'd guess programs that use xlib directly would be the exception. The handful that I've read through don't.

Re:Teacher (aka non-commercial Tutor) here.

[mcrbids]

Unlike VNC which essentially takes a picture of you desktop and sends that, so its easy to see why NX works so much better.

Except that VNC is cross platform. I can see a 'doze system on my Fedora Workstation and I can monitor processes on my Fedora workstation from a 'doze laptop.

Give me that portability with a lower bandwidth constraint, and I'm a convert. Portability is a deal-breaker.

For now, I use VNC, and for Windows-Windows connectivity, I use TightVnc. Usable over a modem, and nice over broadband. Even with a VPN (to enable remote desktop from anywhere, a la VMWare) it's decent almost anyplace.

Re:Teacher (aka non-commercial Tutor) here.

[Rich]

Yes, it's very rare. The other places I do it are code for task management like kasbar.

Re:Teacher (aka non-commercial Tutor) here.

[Bert64]

There are X servers for doze on which you can run apps from a unix system.
As for monitoring processes you can use a text terminal, or just run an xterm inside of an X server, that way you have only one window and not a full desktop...
VNC on unix is POINTLESS... native X11 is faster and more flexible.. Infact, it's rather pointless on windows too, RDP is supported on modern versions and is much faster than vnc, and there are rdp clients for unix.

Re:Teacher (aka non-commercial Tutor) here.

[Bert64]

That's a fault of the toolkits in use, a lot of older X11 apps were perfectly useable over dialup and very quick on dsl speed connections, i remember running netscape in college over a 2mb line, and i did so because the unix server i had access to that was 150 miles away was MUCH faster than the workstations on which netscape was terribly slow.

Re:Teacher (aka non-commercial Tutor) here.

[eean]

Um, hello? Maybe try RTFA. NoMachine has a Windows server. Obviously it can't be as good as it is on X server since he can't use all of its compression techniques, though I understand that its still pretty good. But the security features alone (integrated SSH) make it better for Windows to Windows then VNC... I haven't used it the server end for Windows myself though and its admitedly not their target audience. Actually, Remote Desktop I would imagine to be better for Windows to Windows then VNC, there are clients for it on Linux.

Re:Teacher (aka non-commercial Tutor) here.

[misleb]

So please tell me why X is dog slow while Windows' Remote Desktop is so fast its as if you are using the computer you are accessing directly (it even gives you sound and full keyboard controls).

Microsoft's protocol actually transmits the widget commands, whereas, X only transmits the primitive drawing commands. When you have toolkits like Gnome and KDE that run on top of X, you lose any advantage because their widgets have to be drawn primitively, even though a faster approach would be to only send the data required to tell what widget to draw (not how to draw it). That's why KDE's improvements are much needed, to catch up with Microsoft.

The point was that X was doing it first.

The advantage of transmitting a single application is negligable due to the piss-poor performance of X for any size window.

The performance isn't bad if you have a decent network. And the advantage is still there. It provides much better network transparency and integration. I'll admit that Microsoft has a speed advantage though.

It's much more useful to be able to access a desktop rather than just a single app, especially when breaking the connection closes the single app.

Well, if you run an all Microsoft network, I can see why you'd be so worried about it breaking and losing your connection. ;-)

The Unix model doesn't require that I display very many graphical apps over the network, but when I do need to do it, the functionality is built into the system rather than an addon component as in Microsoft. I can display an application from any unix workstation or server on my desktop without taking over someone elses desktop or worrying about licensing or whatever... even through a firewall. I can easily tunnel X through ssh. With Windows you have to have a dedicated "Terminal Server" and worry about licensing and all kinds crap. As far as I can tell, all Microsoft has is a speed advantage and possibly the ability to play sound and print locally. Beyond that X is far superior in the way it integrates with a network. With improvements like KDE/NX, the Microsoft advantage is all but gone.

-matthew

Re:Teacher (aka non-commercial Tutor) here.

[mcrbids]

Actually, Remote Desktop I would imagine to be better for Windows to Windows then VNC, there are clients for it on Linux

You've just jacked into a DSL line in the lobby of a hotel somewhere in Central California. You have a customer on the phone with a tech support issue. You need to take a look at his computer.

What do you do? You are most certainly behind a NAT firewall - and so is the customer. Do you have a solution for this problem? Well, I do. I bought this back when it was called "Expert VNC". I've used it day in, day out for a year and a half. It's based on OSS software, and I paid $500 for it. It was well worth it. I save that in travel expenses every few days - and it works almost anywhere.

No, I don't get amy $$ for this ringing endorsement, but I do save $$ everytime I use the product.

How nice it is to have customers thousands of miles away who feel as though they've had somebody over the shoulder take care of them! How nice it is, when debugging a problem, to have full access tothe system in question! How nice it is to never wonder what the customer actually sees on the screen!

If $500 is "too expensive" to pay for these kinds of capabilities, you really don't matter, yet. So, "RD" in Windows is just the pits compared to this!

Re:Konquerer, Mozilla, and KMail...

[B5_geek]

You can't use Moz as a file browser.

The example that he stated was to use Konqueror for file management & Moz. for Web Browser.

Re:Not Any Time Soon

[LnxAddct]

I use NX nearly daily. Its just like VNC, just a million times faster and its more clear, so stop whining. NX isn't useless, I've used every desktop program that you can name with it, nothing special needs to be done. I installed the deb with dpkg, added a user, went to work the next day and connected home. Literally nothing else had to be done, it is such an easy setup, I was very impressed. Oh and did I mention, its fast, really fast. I could hardly tell I wasn't sitting at home. Some other cool things you can do with it are printer fowarding and I think something with file transfers, although I haven't used either because I haven't needed to.
Regards,
Steve

diff NX LBX?

[4of12]

How is NX different from the Low Bandwidth X (LBX) extension for the X windowing system that usually underlies KDE and Gnome?

Re:diff NX LBX?

[leandrod]

> How is NX different from the Low Bandwidth X (LBX) extension for the X windowing system

It works.

And BTW, it works with anything X11 too.

Re:diff NX LBX?

[hummassa]

It's MORE low-bandwidth. It caches aggressively on both ends so each side _guesses_ where is the other at that point, and only does a roundtrip to communicate when necessary.

Re:diff NX LBX?

[ADRA]

So does SSH, which is actually the preferred method for low-bandwith consumption by some X developers. LBX was a kludge, we all know it.

SSH tunneling solves the authentication and some of the bandwidth issues involved with X transport. Beyond that, Toolkits need to learn to behave with remoting their desktops. KeithP gave a prety good intro on these issues a long time ago:
http://keithp.com/~keithp/talks/usenix2003/
AND
http://keithp.com/~keithp/talks/lbxpost/inde x.html

Re:diff NX LBX?

[leandrod]

> So does SSH

Not as well as NX.

Perhaps situation will change when everyone is doing vectorial graphics with Cairo sent differentially with XDamage, but for now NX is the way to go.

Mom & Pop solution

[B5_geek]

This sounds like a perfect application to use with Mom & Pop computers.

You can run a proxy server to help filer out all the "Bad Stuff" (TM) on the Internet, and you know it won't be a support nightmare.

Less bandwidth intensive then a terminal session, but less apps too. Might be a good compromise.

Re:How's that different from how things work now??

[KZigurs]

Are you trying to tell me it has better user action response over KDE when I'm sitting at my desktop?

Re:The wierd thing is...

[stratjakt]

I was thinking the same thing.

Seems the answer to make KDE a nice responsive desktop is simple. Get two phone lines, and two modems. Phone yourself, open NX in a gnome session, connect to a KDE session. Hooray!

Re:Not Any Time Soon

[je4d]

Nomachine's NX doesn't need applications or toolkits to be reimplemented to run over it. It works by applying intelligent compression to the X protocol, so any X application can run over it unmodified - this includes KDE, Mozilla, GNOME, whatever.

The KDE integration is only in the kNX client, which takes advantage of KDE technologies such as kwallet, dcop, etc (unlike the 'official' NX client which is pure qt).

Not only MS, but Citrix too...

[Vexler]

An earlier poster replied that Microsoft has had this for years by using RDP. That configuration is not bad, but I would say that the Citrix ICA/IMA architecture has that beat, and more. (ICA/IMA is better at handling burst traffic, and compression is more efficient.)

My company deployed more than twenty-five thin clients in addition to many PC-based virtual sessions that allow the back-end servers to do the number crunching. Each thin client session uses no more than 7-8 Kbps to maintain screen updates, and responsiveness is limited only by the capabilities of the servers and the network bandwidth available.

Re:Destroying NoMachine's business model

[Anonymous Coward]

NoMachine begs to differ as they actually helped in the developement of FreeNX.

Going Blind

[vertaxis]

Click this link to read this thread and avoid the headaches

http://shit.slashdot.org/article.pl?sid=04/08/31/1 745246&threshold=-tid=121&tid=189&tid= 95

Re:Going Blind

[Dr. Manhattan]

...or just set your Slashdot preferences to "Light". As a bonus things will load much faster.

No it is not.

[SlashdotTroll]

KDE is not a library limited by X11 and is slowly revealing that it can be a Desktop replacement for X11 because it can be used with X11. KDE does not efficiently use X11. X11's transport protocol through X11 is only providing the primitives abstracted by KDE's widgets. The only popular credential to KDE is that it can provide a low-bandwidth complete remote desktop with KDE Drag'N'Drop through a low-bandwith duplex network session while having X11-aware syntax. I've been a Xlib programmer for the passed 4.5 years and protocol-efficiency of X11 is only exhibited by Motif. KDE has all the purpose in a Desktop, yet there are many users that find using proprietary widgets is difficult to switch between context with Xlib. Load the X packet session effeciently is more efficiently accomplished when using strait Xlib. I hope KDE, and not either GNOME, ever try to legislate core changes to X protocol and they should stay as X module extensions.

Bullshit

[melted]

Try it. You'll be surprised how zippy it is, even over analog modem. Contrary to popular belief it DOES NOT just send screen captures.

Re:Bullshit

[Com2Kid]

• Try it. You'll be surprised how zippy it is,

Dirt slow, 256kbit up connection, DIRT SLOW.

Click, 2-3 second delay before anything happens, 256 color mode.

(this is connecting to a WinXP box that is using the "Classic" theme.)

Re:Bullshit

[ADRA]

I can't say much against RDP. Its got a decent bandwidth usage, I've heard good things about Citrix more-so since they basically invented Remote Windows. The performance could be better, but it is sufficiant for most operations.

Negatives:

Client (TS Server) footprints are heavier which means you need a lot of hardware to support users. Citrix is good in that if you just need an app here and an app there, you can choose to just export the application, and not the entire virtual screen.

Licensing is and will forever doom TS to an obsurity even in windows. The price point is unrealistic for home/small business and even medium sized businesses will likely only use it sparcely. Ditto for ICA, but i'd say only large companies would use ICA. Its a indispensible feature that fillers like VNC and PC Anywhere will likely never be able fill.

PS: XP bans all foreign windowing tools that aren't licensed, aka RDP/ICA. I love that MS doesn't use its monopoly powers for evil :-)

Imagane an OpenMosix cluster of these.

[LWATCDR]

No really if it could be used like a terminal sever with an openMosix cluster it could be very nice. You need more power you just add more CPUs to the cluster.

Re:Imagane an OpenMosix cluster of these.

[stratjakt]

Mosix won't migrate any processes that access local files, share memory, etc.. So in practicality a mosix-based application server wouldn't be terribly useful.

It would be pretty cool, though.

This Statement Is Not Supported By The Article

[Master of Transhuman]

"They claim to be able to cram a fullscreen KDE session -- KMail for mailing, Konqueror for file management, Mozilla for web browsing and OpenOffice for word processing -- into a 40 KBit/sec modem connection without losing responsiveness for the user experience."

No, they do NOT. The interviewed persons state that a responsive NX session requires a 40kbps link, and about 25MB of RAM. This allows you to run a KDE session remotely and also allows non-KDE apps like Open Office to run remotely.

They do NOT say that you can cram ALL of those programs SIMULTANEOUSLY INTERACTING into that 40kbps.

Obviously they mean you can interact with all of those programs over that link - one program at a time, switching between programs, just like any other remote-control software.

They estimate that a modern PC with 1GB of RAM and a 3GHz CPU could support 25 simultaneous fullscreen KDE remote sessions, crapping out at 35 sessions.

As for usefulness of this technology, they list at least nine scenarios and benefits of using it.

One of which is that it eases Linux adoption on the desktop by allowing Linux clients to access Windows apps running on Windows servers and vice versa, thereby allowing companies to migrate from Windows to Linux at their own pace and not forcing them to find equivalent Linux programs for various Windows-only mission-critical programs. In other words, migration doesn't have to be all or nothing.

Is this too hard for ./'ers to comprehend?

Re:This Statement Is Not Supported By The Article

[archeopterix]

Is this too hard for ./'ers to comprehend?

Short Answer: Yes.

Re:This Statement Is Not Supported By The Article

[gardyloo]

Huh?

Re:This Statement Is Not Supported By The Article

[gardyloo]

Is this too hard for ./'ers to comprehend?

Huh?

Re:This Statement Is Not Supported By The Article

[manyoso]

Please point out in the article where it says anything like this:

"One of which is that it eases Linux adoption on the desktop by {NX} allow(s) Linux clients to access Windows apps running on Windows servers ..."

I ask because this would be a stupendous feat given that no NX server for windows exists. Perhaps you should read/research before berating others.

Re:This Statement Is Not Supported By The Article

[joshmccormack]

One of which is that it eases Linux adoption on the desktop by allowing Linux clients to access Windows apps running on Windows servers...

I didn't see this. They mention a Windows client, but that allows you to run the remote Linux/KDE apps, right? Nice if you want to experiment with this, make transition easier, but not, as far as I can tell, as a way for Linux folks to run Windows apps. Unless the server was running vmware, I suppose, which would be cool.

Re:This Statement Is Not Supported By The Article

[Master of Transhuman]

His statement does NOT say that all of them can run simultaneously. English is either not your strong suit or his. His statement explicitly states that non-KDE apps CAN BE USED inside the session, INCLUDING ANY of the following. This is NOT the same as saying that ALL of them can be.

As to your second point, if in fact ALL of them CAN be so included, that is very nice, but you need to be specific about what you mean when you say "simultaneously". If you mean that you can have windows into all these programs open on the desktop at once, and switch to one or the other of them, that is NOT the same as saying that they are ALL refreshing at the same time.

The headline IMPLIED that the article STATED that ALL of these programs can squeeze their refreshes into the SAME 40kbps data stream. The article does NOT in fact state that, as I pointed out. However, I would be willing to believe it if someone would clearly point to an article or demo that demonstrates such.

If this works...

[HangingChad]

...it's a Windows killer. I know, I know people have been saying that for years. But look, at the pace of development and innovation in Linux v Windows. Okay, maybe a killer, but it will most definitely be a nudge toward nitch market status for Windows.

Now that Shorthorn is starting to look like XP Rebloated, 5% of companies are contemplating a complete switch to Linux and 36% are considering some type of OSS introduction, this could push quite a few more over the edge.

Great idea.

Re:If this works...

[fitten]

Why would this be a Windows killer? These solutions have been around for years on Windows (and in X itself). I can't really see where all the "innovation" is. Five years ago I was administering Windows database servers remotely from almost 2500 miles away with no problems whatsoever using various methods (Windows Terminal Server, PCAnywhere, and now there is another web based one that is even more responsive called gotomypc that we've started using).

Re:If this works...

[bluekanoodle]

I'm curious, where did you get those numbers from?

Re:It's a "thin client"; it replaces nothing.

[Rei]

There are other advantages to the centralized-server model; the biggest one being that you really only need to maintain one box.

Previous Slashdot article

[Krunch]

Here is a link to a previous Slashdot article about FreeNX (without the IT theme).

Replacing Thin Clients

[Bravo_Two_Zero]

It's interesting, but until there are thin clients (no, not the roll-your-own-with-old-PCs variety) that support the protocol, it's a hard sell in a lot of environments. M$ Terminal Services is a pain, and isn't cheap, but we can deploy thin clients with ready RDP sessions in addition to VT220 and tn5250 emulation (including passthru printing). I could do all that with a PC running *nix, but the PC hardware isn't a book-sized device that churns away happily in a dusty warehouse.

It's not a knock by any means. I'd love to centralize the client apps and just serve sessions over 40kbps. But even that is a little expensive over a 128/256kbps frame relay connection. It's nice. It will be useful. But doesn't sound like something to adopt in a real, low-bandwidth, network computing environment at the moment. I'd love to hear that I'm wrong and missing the point, because I'd *love* to replace MS WTS as well as local PCs in our warehouses.

Re:Replacing Thin Clients

[electroniceric]

I'm a little confused by the difficulties you see in incrementally deploying a thin-client solution on an office with existing PC hardware. I'm also not sure what your requirements for your thin clients, but I'd be interested to find out.

To my mind one of the most appealing things this server offers (with the addition of transparent Windows client software) allows is a gradual transition from a desktop Windows/MSOffice environment to a Linux/KDE/OOo environment, whether thin client, or desktop or both. By doing some mapping of /home/~user in the KDE session and My Documents in Windows, you can have both pointing at the same location. And because KDE/OOo/Linux have approximately zero license costs, you can test it on as many or as few users as makes sense, given the retraining and support efforts that would have to go into such a migration. FreeNX comes in as a way to minimize deployment effort and make sure users are logging in securely, while still not paying hefty license fees for WTS/RDP/each copy of Office that runs over RDP.

Now you've run a thin client shop and I haven't so I'm guessing that you've got some wisdom that I totally don't anticipate, so fire away.

Re:Replacing Thin Clients

[bruthasj]

There are plenty of thin clients out there that run Linux and support RDP / Tarrantula / X11 / vt220 / tn5250 / .. and the list goes on. Integrating this, for these vendors, would be a piece of cake.

Anyone purchasing a WinCE thin client is really severely limiting their choices. With a linux-based thin client, a flash upgrade would get you the NX you need. With WinCE: sorry.

Re:It's a "thin client"; it replaces nothing.

[Master of Transhuman]

While you are probably correct in the general case, in this case the FreeNX tech allows 25 full-screen KDE sessions to be run on a commodity box with 1GB of RAM and a 3GHz CPU.

This is not the same as running forty commodity boxes and a "megabrute server".

This is the same as running 26 commodity boxes and supporting 25 users on one more commodity box.

Do the math.

No one is saying this is going to replace desktop PC's. There are specific places where this tech would be very useful (library patron PC's used to access the Net or shop-floor PC's, for example.) Thin clients have their uses.

I agree that thin clients have been overhyped by people like Ellison, but this is still useful tech. As the developers suggest, this can aid migration from Windows to Linux by allowing companies to run mission-critical Windows-only software on their current Windows servers and allow Linux clients to access it - and vice versa.

Re:Not Any Time Soon

[asoap]

You forgot NX - Nitrous Express.

Really it should be NE, but that's not as Xciting.

-Derek

Re:How's that different from how things work now??

[Progoth]

What I would call an archievement would be a KDE GUI that wouldn't feel like being VLC'ed over 40kb/s modem connection on my p4-2.8/512/MatroxG400DH desktop.

I don't know what's wrong with your computer, my athxp2500+/512/gf4ti4600 runs kde3 a whooole lot faster and more responsive than winxp.

perhaps that is a crappy video card (I've never met a good matrox although i've heard they're out there) or you're running redhat/fedora.

Another entry in the central vs. distributed power

[CFD339]

...fight that's brewing. This may be the linux entry.

IBM is pushing toward a new (centrally managed and provisioned) eclipse based rich client (aka browser on roids) to talk mostly to its workplace portal (aka websphere with portlets)

Microsoft is pushing the other way with a more complex workstation model and an enhanced "user controlled" sharing place via sharepoint portal,

Linux desktops have, generally, been trying to mimick Microsoft stuff but with secure, stable applications....(e.g. open office)

Now, a KDE may become a player in the rich but centrally managed client space.

Interesting.

Clearly there is a place for centrally managed rich client -- think live of business stuff that at one time was managed through the 3270 terminal. The browser has proven it sucks for that kind of thing, but this trend toward a semi-open or open rich client replacement is going to be the thing to watch.

Line workers are going to use this stuff (IBM's or something like it -- this KDE version could be it too). The big battle will be for how corporations manage knowledge worker desktops. Will it be a Centrally controlled rich client, or a traditional powerful workstation with sharing tacked on.

It seems like Linux based machines are likely to have a horse in both races.

How does it work?

[RAMMS EIN]

I tried to search for an explanation of how NX works, but it didn't turn up anything useful. Can anyone explain how NX works?

Re:How does it work?

[be-fan]

NX is a compression and caching layer on top of the X11 protocol. It takes the basic X protocol and performs compression of protocol requests, caching of server responses, compression of images, etc.

Re:How does it work?

[Abcd1234]

It works at the X protocol, rather than frame buffer level, which allows it to perform optimizations that VNC cannot. It also doesn't require a special graphical client... NX acts as an X protocol proxy, so your remote app displays locally like any other X application (as opposed to being contained in a separate desktop within a VNC session).

Re:How does it work?

[Alien Being]

here

Re:How does it work?

[xybe]

From their site: NoMachine has developed exclusive X protocol compression techniques and an integrated set of proxy agents that make it possible to run complete remote desktop sessions, even at full screen, using narrowband Internet connections, at speeds as low as those offered by a 9600 band modem.

They use X at the base and add further compression and network optimization. In fact they make it work on top of VNC and RDP

Not too shabby

Re:How does it work?

[TeknoHog]

I thought they would be doing something KDE-specific to compress even more. For example, if the QT library were an extension of X, you wouldn't have to send the widgets to the server as bitmaps.

On the other hand, if this is not limited to KDE, it's even better :)

Adding to my own post:Replacing Thin Clients

[Bravo_Two_Zero]

I should clarify. I have seen this:

RDP and RFB Foreign Protocols

NX accessibility and remote computing capabilities are not limited to Linux desktops and servers. NX encapsulates and translates into X protocol the Remote Desktop Protocol used by Microsoft Windows NT/2000 Terminal Server Edition and Citrix Metaframe, and Remote Frame Buffer, the protocol used by VNC, another Open Source remote computing facility, available numerous different operating systems.

Although NX compression offers the best performances when running native X applications, RDP and RFB sessions can be compressed by a factor ranging from 2 to 10. NX support of foreign protocols provides further advantages. Firstly, it extends its reach to virtually any computer and secondly, NX offers to the user a unified view of any application resource available over the Internet. ...so I might be a total darsh. But I still haven't been able to find an answer on the nomachine.com site as to whether that means it works with RDP clients (I think likely not), or that it serves RDP sessions via the X protocol (yeah, that's my guess).

Re:Two words...

[BrookHarty]

bull...shit

Remote desktop is faster than TightVNC over a modem, but TightVNC just uses enhanced compression compared to VNC. LBX doesn't work over large latency high bandwidth links.

I've always suspected that if someone took the time to figure out how to do X-windows with better compression and client-server model, you could compress it down to a much better experience.

Back in the good ole text days, people wanted to do RIP grafix, and started using 64 or 128 bit numbers to pattern match and speed up graphics. Then there was a time people tried to use fractal compression to speed up photos over the web. (OR something even older, RLE graphics over 300 baud)

There are many methods that can be expanded and combined to make a low bandwidth, high response session. Maybe trading color for speed, a monochrome experience that is fast is better than a slow true color display.

Saying it can't be done is incorrect, there are many methods. Not sure if todays programmers even know of these methods, or have experience in low bandwidth situations.

It can be done, will it? With people getting more high speed connections, optimizing is a thing of the past. Same goes for video games, they come on DVD's and/or multiple CD's. The only place you see true optimization is the demo parties, 64K demos with extreme amounts of content. 64K would stream over a 28.8 modem in a flash.

My idea was use a server-client model, using content updates with a highly compressed pixel line in 1 byte patterns, widgets are cached, and text streams are compressed.

Not exactly sure how NX is speeding up RDP, other then widget caching, the articles appear to be slashdotted already.

Re:It's a "thin client"; it replaces nothing.

[MindStalker]

NX isn't so much thin client as it is remote desktop over slow connection. Think connecting from home or other slow connection to your work computer and give it jobs to do. While this application may lower bandwidth for existing thin clients. Its not the real drawl.

Re:where's the low-cost terminal?

[stratjakt]

Well, how much for a nano-itx board, external-modem sized case, wall-wart PSU, handful of RAM - a thin client shouldn't need much, and a CompactFlash-IDE boot drive? Or can you boot a nITX board over the 'net?

I can't imagine something like that costing too much if it was assembled for the mass market, and it meets all the specs you want. I'm not sure how many watts they suck down, but it can't be much.

Re:How's that different from how things work now??

[chill]

Network clients are the future. Things like NX, JDS, Citrix and the like are massivly deployed in the corporate world.

As an administrator, I love the centralized configuration. The ability of a person to move from their office to a conference room and pop right back into a session is fantastic. Add the ability to enter a session from home or a VPN tunnel and it is really useful

In the office, that means many people can use inexpensive PC's or thin clients and get everything but 3D. Maintenance is a breeze, since I don't have to install whole bundles of software per machine or tweak up Ghost images it saves me time.

For most of my users, it makes things easier. They can access their work and stuff from anywhere. The network is either switched 100-base or in some cases switched 1000-base, so there is NO LAG in loading/saving data or running programs. Hell, stuff loads/saves FASTER through Gig-E to our SAN than to a desktop IDE drive!

An NX Server behind the firewall that I can tunnel to gives me a gateway to every PC in the building via translated RDP (Windows) or VNC. AND it seamlessly encrypts the sessions (unlike VNC). All I need is one hole in the firewall for the NX server, instead of one per VNC box. (Yes, I could tunnel VNC over SSH or stunnel, but that is a pain in the ass and NX is so much easier.)

And if KDE 3.2.3 or 3.3 is slow on your stated config, then something is wrong. Spend an evening and compile your own version w/P4 optimizations and remove the excess items you don't use. It should scream.

Ooh, aah

[Anonymous Coward]

...into a 40 KBit/sec modem connection without losing responsiveness for the user experience.

1996 just called to say "Thank you".

Re:Two words...

[Com2Kid]

Programmers now tends tend to, unfortunately, use the "Code it using the most braindead algorithms, and rely on bandwidth/CPU/memory increasing over time."

Software isn't made to be usable at time of release, it is made to be usable two years after release.

Re:How's that different from how things work now??

[Com2Kid]

• perhaps that is a crappy video card (I've never met a good matrox although i've heard they're out there) or you're running redhat/fedora.

I take exception to that, Matrox video cards kick ass in the 2D arena, and are built damn well. I cannot speak for their *Nix drivers, but their Windows drivers are rock solid stable.

Kick ass work horse cards that sit down and do the job.

Windows servers?

[loqi]

TFA < /.

Just curious, did they mention any Windows compatibility with this? Unless someone writes some kind of translation layer for Windows API calls (which sounds like a lot of work), how does this assist migration from Windows apps?

Re:Windows servers?

[hummassa]

FreeNX is not really Win compatible, but NoMachine's NX Server has a Win version.

Belongs in SSH

[Effugas]

NX really needs to be ported to into OpenSSH as an optional compression module for its X Forwarding component. That way, there's literally nothing more than:

ssh -X user@host ...and if both client and server support NX, things just fly.

--Dan

Re:Belongs in SSH

[Effugas]

Yeah, but this is OpenNX encapsulating SSH, rather than SSH encapsulating OpenNX. The latter is, architecturally, the most simple and straightforward way to deploy NX.

--Dan

Re:Belongs in SSH

[agoliveira]

That way, there's literally nothing more than:

ssh -X user@host ...and if both client and server support NX, things just fly.
Sorry but it's much more than that.
One important thing is the virtual elimination of roud trips which ssh alone can't do.
This does *a lot* of difference alone.
Besides that there's differential compression, double side caching, etc.

Re:Belongs in SSH

[Effugas]

OpenNX operates on X streams; I don't see why things would be any different if you operated inside of SSH instead of being a front end to it. If I'm wrong, please mail me, dan@doxpara.com .

--Dan

Licensing Problem

[Anonymous Coward]

And even then, you have get a single "stock" license, so no more than one connection at a time unless you shell out some bucks.

I have a problem. Even though I constantly used TS on a Win2k server it keeps issuing me a new 90 day temporary license ever 30 days, thus never requiring a regular MS TS CAL. Here's how it happenned:

1. Installed rdesktop
2. Created the following script to run rdesktop:

shortdate=`date +%y-%m`
rdesktop -n $HOSTNAME-$shortdate w2kserver.local

I guess I'll figure out the problem someday. Meanwhile my list of temp licences is getting long. Please don't take advantage of exploit.

Grammar nazi alert

[sean.peters]

Are they inferring that...

No, %$#@ it, they're not. They might be IMPLYING it, though.

For God's sake, people. The difference between "infer" and "imply" is not that complicated. We ought to be able to get it straight.

Sean

Re:Grammar nazi alert

[cpeterso]

You should just give up now. You're fighting a loosing battle..

Re:Grammar nazi alert

[rvega]

You're fighting a loosing battle..

It's a %$#@ losing battle.

But at least you wrote you're instead of your, so perhaps the battle isn't hopeless after all.

Re:Grammar nazi alert

[Scaba]

It is acceptable to use "infer" in place of "imply".

Re:Grammar nazi alert

[Espectr0]

4 entries found for infer.

1. To conclude from evidence or premises.
   
2. To reason from circumstance; surmise: We can infer that his motive in publishing the diary was less than honorable.
   
3. To lead to as a consequence or conclusion: "Socrates argued that a statue inferred the existence of a sculptor" (Academy).
   
4. To hint; imply.
   

Re:Grammar nazi alert

[julesh]

What's your source for that? Concise Oxford mentions your definitions 3 and 4 but warns that they are disputed (i.e., you should normally avoid using them because many people won't understand what you meant).

Anyone else remember GraphOn terminals?

[PythonCodr]

... or am I the only dinosaur who remembers these?

GraphOn made this really sweet line of X terminals that allowed you to split the X server between the remote workstation server and the the display/mouse/keyboard. I was lucky enough to have one of these at home, and it was very zippy ... at 9600 baud I could run an X display that was darned nice to have a full X display at home while my VaxstationII sat at work. Later versions used better compression and were even faster and more responsive. They used all sorts of tricks with save-unders, display lists, and mouse-overs to keep the actual line traffic as low as possible.

Granted, that was the late 80's, and X was in its infancy and clients weren't as feature rich as they are today (The web? Oh ... that thing that's going to replace archie, wais, and gopher...), but it worked just fine for what we used them for. Even at 2400 baud, you could use 'em, but you really wanted at 5400.

Ah ... those were the days, when you could have a 12" X display at home ... 17" if you were really, really lucky. And they screamed at 19.2k! :-D

Re:Anyone else remember GraphOn terminals?

[codepunk]

We buy neoware capio one machines now that run redhat with remote x capabilities. The price is outstanding and it can hook to anything plus a kitchen sink.

fanboy alert

[slittle]

Isn't PCAnywhere like VNC? So does X11 allow you to detach an entire desktop environment and reconnect to it later, with all the apps within still running like nothing ever happened?

There are different uses for roaming desktops vs exporting an application's display.

Re:Konquerer, Mozilla, and KMail...

[stuness]

I find Konqueror underperforms (compared to Mozilla) in some web-browsing aspects.
For example, layout can be a problem. But I use Konqueror almost exclusively anyway.
When I run into a problem, I simply use the "Location"->"Open with Mozilla" command.

I do not use a file browser locally, but often navigate directories or look at source code remotely.
The rich feature set (e.g. syntactical highlighting of source code, support for smb) won me over.

Re:How's that different from how things work now??

[codepunk]

No and that is not the intent, nx is x protocol compression that increases the efficiency and responsiveness of applications over slow high latency connections.

Re:Two words...

[cyb97]

to be technically correct, a 64k would stream over a 28.8 just as slow as anything else...

Firstly the modem doesn't get faster because the content is smaller, and secondly streaming sort of implies just showing output and not copying the actual executable.
A 64k demo is just as intensive as any other program in the same resolution/refresh...

Re:Two words...

[the chao goes mu]

And the trends tend to support this abuse of bandwidth. Two examples:

XML instead of fixed width or delimited data

Serialized Java objects instead of simple data structures

(Don't flame me about either XML or Java, as I know they have proper uses, but they are not the proper solution for every situation as the buzz-word worshippers would have one believe. Using a serialized object to carry a single string --which I have seen-- is a meaningless waste of bandwidth. So is replacing 4 lines of SQL with 120+ lines of XML, which one of my employers tried to do as well....)

Yeah E17 IS GONNA BE GREAT!!!!

[7-Vodka]

Maybe when they release it, my grandkids will be able to use it along with duke nukem forever and the next debian stable release. It will do well to compete with the successor to longhorn also.
Good job E guys.

I keed, I keed.

But in all honesty, I'd like to see some people get paid to work on E17. Because then we might actually see someone who gives a flying fuck about the users.
Being a user myself, it sucks to get the attitude of "oh well, you're just a user. Of course you are not worthy enough to actually be able to taste of this E17, which btw is the broth of the gods 100000x better than anything you've ever used.

NX is to VNC

[Julian Morrison]

...as X is to frame-buffer.

VNC runs an app remotely, displays it remotely, and sends bitmap movie of the display actoss the network. It can't scale, because the server has to do 100% of the work, and because sending bitmap diffs is bandwidth heavy.

NX runs an app remotely, displays it locally. Only the unavoidable parts of X protocol travel over the network. It can scale well, because the server only does the bit-crunching; the "thin client" draws the display.

Re:NX is to VNC

[Guspaz]

Sorry if this is a stupid question, but how does NX differ from XDMCP? I'm not too clear on that.

Re:NX is to VNC

[yuri benjamin]

I think NX is more like xmove, except it works.
xmove appears to be an abandoned project.

ob /.

[5m477m4n]

from TFA:

imagine a blade center as offered by HP or IBM, to allow several hundreds of parallel sessions.

I couldn't have said it better myself.

Re:It's a "thin client"; it replaces nothing.

[bluekanoodle]

Not quite. Don't forget the TCO. It's alot easier and quicker to maintain and update a a properly setup thin client soluton serving forty nodes then 40 commodity boxes.

I switched my company over to citrix and we ended up saving alot.

1) I could use our legacy pc's in a locked down state as clients. This avoided having to buy new pc's just because our accounting app needed a faster processor.

2)with centralized administration, we were able to avoid having to hire another staffer to handle support calls.

3) When a piece of hardware dies, I can replaced it with a QUALITY thin client appliance for a less then it would cost for a QUALITY commodity box. Sure I could buy cheaper no name hardware, but I wouldn't stake my job on it.

4) Our customized software does not need to be rewritten for different platforms. Doesn't matter if the client is running Windows, OS X, Linux or an embedded OS. they work exactly the same on each platfom. Not kind of the same, not sort of the same, but exactly the same. This saves on training the monkeys, I mean end users.

We can also provide secure remote access to our data without worrying about whose using what license, and whether their offsite machine is compromised.

At our current growth rate, we save almost 40% with thin clients over commodity boxes. That's not some number pulled from a marketing whitepaper, that's an apple to apples comparison from our department budget when we looked at both scenarios.

NX works now

[rRaminrodt]

NX works now, with the existing X11 apps you already have. That doesn't just include KDE and GNOME, that includes xclock and emacs and acrobat and so on... that includes those oddball legacy X11 programs that you can't reasonably rebuild to support new libraries. I think that's a pretty valuable ability.

Re:How's that different from how things work now??

[JamesTRexx]

Actually, whether it's unix or Windows, networked clients are very much alive nowadays.
For instance, the company I work for has one division with a headquarters in one town, and 11 smaller offices scattered across the country. At first they each had their own server and network, now all they have are mostly citrix clients and a connection to our main network. It saves us a load on support and troubleshooting problems with users running applications. It might be somewhat slower than straight from the desktop, but there is no loss on productivity. (unless the connection is breaking down *cough*damnedversatel*cough*)
Giving everyone a fullblown pc with every bit of software on it isn't always a good thing. Some things work better that way, some work better in a client/server environment, and it'll be even better if we can do it in a secure, stable environment without all the usual Windows hassles. (even with extensive security measures)
btw, I run KDE on FreeBSD on a Celeron 700 with a Matrox G200 videocard without problems.

a good start

[5m477m4n]

Per running KDE user session, an NX server takes about 40MByte of RAM and 100Mhz of CPU. A current standard PC as sold today, with 1 GB of RAM and 3Ghz CPU should allow 25 sessions in parallel without any problems. It would probably get flanky at 35 parallel sessions.

I'm as pro Linux as the next /.'er, but this isn't all that impressive. I maintain a Citrix server a little less powerful than that and it supports over 100 sessions easily. But it is still a good start.

Re:No, wrong idea

[Anonymous Coward]

Can we please let go of the dream of using X, NX, VNC or anything else over dial-up or slow links? It's not happening people.

Shutup, you wiseacre of a Don Quichotte! Shutup at least until you have testdriven NX.

I've used X and VNC over 100Mbit LAN and 802.11b. It's useable but frustratingly slow.

OK -- at least here you are right. Continue with that part.

To give you an answer to your real question::

• it is so slow because of the excessive roundtrips modern toolkits and GUIs load upon the marvellous X11-protocol
• NX is so fast because it gets rid of these roundtrips altogether

Now go do your homework. Testdrive NX and then come back to report.

NOT KDE

[ajs]

Let's be clear. This is not KDE. This has nothing to do with KDE, any more than KDE having an AIM client ties AIM to KDE.

NX is not toolkit-specific, it's just a way of compressing the X protocol for displaying applications over low bandwith connections.

That said, the KDE folks are talking about "integrating NX" into their KDE application framework, which would presumably mean having desktop tools that make the use of NX more convinient, and perhaps wrapping some of KDE's out-of-band data into the NX protocol (such as inter-application communication).

This is all good, but people are missing the mark if they think this is a special way of moving KDE (that is, Qt) widgets across the wire. It's simply not.

indeed!

[jeif1k]

Evas-based apps are at least 10 times more responsive than GTK/QT apps That's not surprising: Gtk and Qt make horrendously poor use of the X protocol, memory, and resources. Performance problems people claim to experience with Gnome and KDE are due to the toolkits, not X11. Of course, machines have become so fast that even Gtk's and Qt's inefficient use of X11 doesn't make any significant difference over modern networks and on modern machines.

Re:indeed!

[Bert64]

Perhaps a project should be started to improve the efficiency of these commonly used toolkits.. Older X11 apps were useable over a 128k line, now i have 512kb dsl and modern apps are useless...
With widespread broadband X11 apps should be awesome over the network, instead theyre even slower than they were back in the days of dialup!

[OT]Re:I don't believe Flamebait???

[hauer]

What's going on here, please? Allright the guy maybe wrong, maybe very wrong but there is ten times more thought (call it insight) in this post than what I am used to see here on /.

If I called anything flamebait, it would be the two replies whose authors' vocabulary is not rich enough for a decent and civilized phrasing of "you are mistaken".

Re:It's a "thin client"; it replaces nothing.

[ClosedSource]

Your number 3 point suggests that your QUALITY commodity box can fail without you losing your job. It sounds like you're making excuses for the relative overexpense of thin clients.

remote intrusion---echelon/carnivore capabilities?

[davidsyes]

This is neat/fearsome/interesting because just a few minutes ago as I was reading:

http://www.securityfocus.com/infocus/1769

my mind wandered (like it always does) and I asked myself a question I'd pondered many times:

"How'd the FBI get the password from that mob/mafia guy they nabbed? They refused to tell the public how they did it."

We'll I USED to suspect they used keystroking, whether they paid an insider to put some gizmo into or next to keyboard to pick up his keystroke's raw RF. I also suspected they pointed some microwave antenna into his home/unit and had a receiver/ticker/counter on the opposite side to measure differences in power/strength/amplitude/modulation or whatever else they could. (This assumes the target was not marginally smart (or informed) enough to shield his home against RF/microwave (or if fearing Star Trek, IR) snooping.)

JUST as I decided to to check Slashdot to see what is going on, I see this thread/topic. I immediately thought, "I'll be damned, they probably for some time had THIS capability and fed the guy a fake but remote shell of his own session, soaked up all his passwords and notes he thought were local (maybe he/the mobster never gave it much thought).

So, what are the implications for this if a Man-In-the-Middle heist of signals is perpetrated against a perp or a citizen under observation? In school, work, or a controlled environment, it can be expected to be fed a remote session, be monitored, and have no privacy. But imagine if not only crooks and crackers intruding on you, but imagine if domestic intelligence is able to or is actually hijacking entire screen sessions and using them.

But, I imagine there are better ways than this. However, does the remote session leave any forensic dtails that it was in effect? I imagine a number of remote tools would leave evidence, but when a government uses Echelon or Carnivore, I imagine they do it at the ISP level JUST because the session can be recorded at the ISP's equipment, and not likely be noticed on the target's own hardware (after all, mobsters and crooked accountants probably have magnetometer sleuths for friends...).

Just a thought, or two...

David Syes

MS and Citrix were late to the game

[jeif1k]

Network transparent window systems have been in widespread use since the 1980's. RDP and Citrix are the typical "me-too" products in the Windows world, years late and not quite as good (but probably good enough to get the job done). Note that remote desktops are not necessarily aimed at dial-up. For example, X11 was not designed to use the smallest amount of bandwidth, it was designed for a LAN environment and still is far better than either RDP or Citrix on a LAN. Raw X11 is not the best choice for dial-up, but in combination with something like DXPC or NX, it is good there, too. And for the ultimate in thin clients, VNC is a great choice; VNC represents a completely different tradeoff in design parameters for remote displays.

can you disconnect/move sessions with NX?

[jeif1k]

The biggest problem with remote access using X11 in the past hasn't been performance, it has been the difficulties of disconnecting/reconnecting applications. Originally, this had been envisioned as an integral part of using X11, but it was the applications that were responsible for supporting it and they never did. Eventually, we got tools like "xmove", but they were too cumbersome for general use. VNC succeeded so well probably not because of its compression (which is decent; in particular, it works with ill-behaved X11 apps like KDE and Gnome desktop apps), but because it made disconnecting and reconnecting sessions so easy.

My experience with NX...

[chrispyman]

Last week actually I tried out NX server as it came on the new Knoppix cd. Compared to the alternative I once used (VNC), it's really fast and responsive. Infact it's just as good, if not better than Windows Terminal Services, which was suprisingly responsive under 56k. Just as a real world test, I hooked up the Knoppix NX server to my DSL line and then connected via Dialup on my laptop and It honestly felt like I was using a slightly laggy Linux workstation as opposed to a horribly laggy one that it was when I tried VNC. All in all, I'd say that NX is gonna give VNC, as well as Windows Terminal Services a run for it's money.

Emphasizing my comments. I am not a troll

[Anonymous Coward]

Rich,

If I had a nickel for how many times I need to admonish others that I am not trolling, I would donate all those nickels to X.org.

First, KDE needs Xlib to connect to an X Server. KDE has graphics paths to Microsoft's GDI as well as X11. KDE doesn't have a rendering path directly to anythin outside of Microsoft's GDI. What I was trying to say is that KDE (and the same to be said for Gnome) is limited on Linux environments by the X Server. What you and I will slowly see is a reversal of technologies by the dominant *Desktop environments to establish a local API to access the canvas (think graphics hardware) directly and not needing to move through Xlib on the network datalink.

The methodology that I am aware has already discussed this has been readily available in the Linux kernel as the "framebuffer" graphics modules and an alternative linux-only implementation known as DirectFB. At the DirectFB project's homepage, there are many sub-projects. DirectFB is the framebuffer drivers where currently GNOME and generaly anything using the GTK+-2.x (and GTK+-1.x Iirc) can use a framebuffer with a driver-based unnetworked window-managing system with window transparancy; ie everything KDE and Gnome Desktop desire without needing to move through Xlib in the final packet session output to the X Server. It is all done without the X Server. Further into the DirectFB project website you will find a X Server that *runs* on the framebuffer; tricky thinking, that I suggest to you that this X Server is interfacing its windowing environment regulated by the DirectFB's natively-encoded window manager; XDirectFB. As much as I dislike linking and using enterntainment software to discuss the innovation in DirectFB's many subprojects; I do reveal to you that there is even the project DirecFBGL demonstrating using the Direct Rendering Infrastructure without a X Server and in the Framebuffer provided by DirectFB. Quake3, accelerated, in a Framebuffer, I apologise for only having posted an example of entertainment.

K NX is the prequel to DirectFB; NX server tries to keep X11 and KDE tightly bonded because KDE has no rendering path through anything but a X Server and Microsoft GDI. I need not say more.

Despite you labeling me as a troll, I forgive you. I've made mistakes, and one of them was not to secure "SlashdotTroll" from the hands of debilitating users to post pornography and slander.

HTH,
-SlashdotTroll

Re:Emphasizing my comments. I am not a troll

[eean]

I still don't understand what your saying. "NX server tries to keep X11 and KDE tightly bonded because KDE has no rendering path through anything but a X Server and Microsoft GDI. I need not say more." is a sentence that does not make sense (the presiding paragraph does, but doesn' seem relavent). Exactly what is your source that NX tries to keep X11 and KDE tightly bonded. And what does that mean regardless, I have an image of the KDE dragon in bondage gear. My point was NX may use KDE GUI for its client, and the FreeNX may be created by KDE developers... but the NX server doesn't have much to do with KDE.

Re:Emphasizing my comments. I am not a troll

[eean]

...could someone point me where it says the NX server is somehow compiling KDE and X together. Thats what my message was all about in the first place, cause I don't think its really the case at all. The NX server doesn't have anything to do with KDE, other then its a possible DE that it can display (and sound server stuff, but thats neither here nor there).

On the SPAM and v1.a.g.r.a subject

[hummassa]

It's funny. 99% of all spam I receive is from the US, not from BR (and I had my work e-mail in a public mailing list, up to harvesters) -- the other 1% is from the Far East, and in a Far East encoding.

NO SPAM at all from Brasil.

Re:It's a "thin client"; it replaces nothing.

[GreyWolf3000]

Plus with Linux you don't pay for the OS that gets installed on each dumb terminal, so it actually does end up being cheaper.

Too many hyperlinks in story

[vlauria]

I hate when people hyperlink every other fucking word in a story they submit. For this post, I don't need links to Mozilla or Open Office to gain more information on the story, it's a fucking nuisence if you ask me.

Re:UNIX has had it even longer

[szo]

and why is that, can you explain?

tia

Rebuttal. Why is that modded up?

[zealotasd]

>>NX takes the X protocol and uses various caching and compression methods to make it more efficient.

I've never read so much arrogance in my life. SlashdotTroll wasn't trolling. You described what SlashdotTroll summarily protested the purpose of the NX server.

And apparently because SlashdotTroll is throwing around a different dialect of English with *many* keyboarding errors, she/he rebutted your argument under an anonymous post because obviously enough people have modded down the post because the word "Troll" appears in the userID.

If SlashdotTroll is a troll, then how did you get so on-topic as did the post you replied to and yet you or someone weilded moderator negation with no more merit than declaring "I'm pretty sure the parent is BS or I just can't read what its saying."

There was a similar story about a Man whom people declared as being the Son of God, and that said man, spoke somthing along the lines of;

(Acts 9:16);
"For I will shew him how great things he must suffer for my name's sake."

NX has been around for awhile.

[Ayanami Rei]

It's just only recently that it's been recognized to exist. It originally started out as DXPC (Differential X Protocol Compression) back in '95, and people were working on similar things earlier than that.

NoMachine started packaging their own implemntation of this concept in an improved, commercial form as NX. About a year ago NoMachine makes the new vesrion open source, but I'm surprised it's been this long and still people don't know anything about it.

Hell, even DXPC could be as good as RDP. But it wasn't easy to set up so...

Re:No, wrong idea

[kermit6306]

it is so slow because of the excessive roundtrips modern toolkits and GUIs load upon the marvellous X11-protocol

Did I stutter? X is slow. End of story. I imagine whatever toolkits you through on top only exacerbate the situation.

Testdrive NX and then come back to report.

Will do. No doubt I'll be left unimpressed, per usual.

Re:Two words...

[BrookHarty]

to be technically correct, a 64k would stream over a 28.8 just as slow as anything else...

We are talking time per response, if you can compress 1024K to 64K, you speed up the response time for interactivity, slower links appear faster. That was my comment.

Firstly the modem doesn't get faster because the content is smaller, and secondly streaming sort of implies just showing output and not copying the actual executable.

WTF? if you have a pipe, and you dont fill it, its fast enough for the need. The idea is to reduce the flow of data.

I have no idea what you are talking about.

Re:Konquerer, Mozilla, and KMail...

[codemachine]

Well you can, it just isn't a very good one.

IDIOT'S BE GONE

[ThoreauHD]

Do you know what a remote desktop is? You can run any and all programs you want fullscreen/thin client Desktop on 40Kb/s. That is huge. Citrix can do the same at 25Kb/s.

There is the same bottleneck at opening up multiple sessions on Citrix as any other thin client tech. Multiple sessions isn't the point. One session at modem bandwidth is the point. You publish a desktop with any and all of your apps on it and you are done.

The difference is this. Citrix is $12,000 per 20 users. This is free. Are you grasping the possibilities now? Is the ./ effect seeping into your brainstem? Either way you don't administer any users- so it doesn't matter. But for those that do, it matters.

Re:IDIOT'S BE GONE

[Master of Transhuman]

Yes, I know what a remote desktop is.

The headline implied that the article stated something specific about performance. Some /.'ers questioned that performance implication. I pointed out that the article itself did not state that all of these programs WERE operating simultaneously in a given NX session. The article states that all of these programs were ALLOWED to run in a given session - which is not the same as stating that they in fact DID ALL RUN in a given session. The implication was that they could, but this was not stated explicitly.

I did NOT state that it is either impossible or even hard to do this. I don't know, since I haven't used the product or Citrix.

I merely pointed out that the headline was misleading.

Apparently most /.'ers flunked simple English.

And I am perfectly well aware of the value of the product if in fact it works as advertised. My other posts in this topic clearly said so.

I would say the /. effect has eaten your brainstem, rather than merely sinking in.

Re:It's a "thin client"; it replaces nothing.

[drinkypoo]

Thin clients are appropriate for everything but high-volume-of-data applications like video. The only problem is that they cost way too much. There is no justification for what amounts to a low-end PC with an outdated video card costing five hundred bucks.

LTSP/ mutliple processors

[denny_d]

Noone has spoken about LTSP. I use it all of the time and I could certainly use technology that would squeeze more client sessions on my aging network. Secondly, is there any mention of how this would work on a multiple processor system? Is there any use of SMP?

Re:LTSP/ mutliple processors

[denny_d]

Skolelinux http://www.skolelinux.org/portal/ will probably be the first to bring NX to the schools that really need this technology *simply* packaged. There is also k12LTSP but they are based on Fedora. Skolelinux is debian based.

bad form but who cares

Thinning out X

[gavindi]

This is good stuff this NX server. I wonder how the new XDamage extension that is going into the Xorg server will help NX or if it will go a long way towards making NX redundant.

NX and LTSP..

[mrhandstand]

might be really interesting. The 100Mbs link to power terminal servers just got fatter in comparison. In fact LTSP is already talking about it I believe. FOr those who haven't checked it out, your should. And think about their work combined with NX...

Screensaver

[owlstead]

Before anyone starts experimenting with this technology....

!!! turn off that screensaver !!!

Some screensaver generate (more or less) random pixels. On any 1024x768x3 screen, run over a network, this will wreak mayor damage.

I've seen this with VNC First of all, your X server will become irrisponsive, and secondly your network will be brought to a grinding halt (time to remove those hubs and place switches guys).

No matter what you do, such data can not be compressed by any means. Only running the program locally will work.

I have to say I'm impressed with NoMachine

[stickystyle]

I connected to there 2Mb test machine over in Germany (im in CA, USA) and thought, 'hey, this isnt that laggey after all'. Then just on a whim i thought to check the uptime...
03:06:25 up 77 days, 16:35, 1543 users, load average: 2.71, 2.31, 2.29
1543 people logged in on a 993MHz box and typing is still not lagging so bad that it upset me.
kudos to NoMachine.

Re:It's a "thin client"; it replaces nothing.

[aminorex]

> While this application may lower bandwidth for existing thin clients. Its not the real drawl.

And as anybody who's been to the great state of Jaw-juh knows, a real drawl will dramatically lower your bandwidth!

Re:It's a "thin client"; it replaces nothing.

[bluekanoodle]

No I'm saying the thin client appliance and quality commodity boxes have a proven history of long term reliability, so I don't have to stand before the cfo and explain to him why the cheap boxes I bought last year are are failing at an EXCESSIVE rate. The Thin Clients we buy now have no moving parts so mechanical failures on the client end are almost unheard.

Re:How's that different from how things work now??

[Tyreth]

" Good people do not need laws to tell them to act responsibly, while bad people will find a way around the laws. "

Same argument could be used about locks on cars and houses. It's not that the law stops bad people, it just hinders them. But, obviously, some laws are useless in regards to helping the good and slowing the bad.

A post that makes me want to know...

[aminorex]

Where is the +1 obscenity menu item?

I've got a cool use of NX technology.

[Linuxathome]

I realize that this may sound like a commercial, but for the overall price of this setup, it's so enticing to get started ($20USD a month). You can get virtual Linux hosting at rimuhosting.com and essentially have a linux box online all the time, accessible anytime you want with 3GB of disk space. If you want a usable desktop anywhere you go in the world that you want to access remotely--reasons for doing so can be one of many, such as accessing a remote server to run a browser to access content restricted to a certain IP, etc.--you would just need to install X and, if you wanted, one of the desktop managers (KDE, Gnome) and then the FreeNX server, and run the NX client to access the box. I have to say that before with VNC, it was next to impossible for me to do so because of the sloooowwww responsiveness, but NX is so much faster. Granted, you have to run a *Nix box running X (no other desktops are supported as in VNC), but the ability to access a desktop remotely anywhere in the world sounds awesome. Now if only I can think of why doing this would be worthwhile.

Re:Rebuttal. Why is that modded up?

[eean]

You (and perhaps the moderators) don't understand the difference between a troll and BS. I didn't really think SlashdotTroll was trying to elicit some heated response, the message just didn't make much sense and it seemed to have factual errors, though it was kind of hard to tell.

Re:How's that different from how things work now??

[chill]

Same argument could be used about locks on cars and houses. It's not that the law stops bad people, it just hinders them. But, obviously, some laws are useless in regards to helping the good and slowing the bad.

Not quite. A law doesn't hinder someone, unless a cop happens to be standing right there. A law is a guideline w/consequences whereas a lock is a physical deterrent.

Do you not steal because there is a law, or because it it wrong?

My sig is targeted at the "there ought to be a law!" mentality. Most people never stop to think that there probably already *IS* a law, or the law wouldn't have prevented anything anyway. Just remember, the latest statistics from the US Department of Justice shows that we have over two million people now locked up in American jails. There were 702 prisoners per 100,000 population. (More than Russia, Iran or Germany.)

For each of these there was ALREADY a law that didn't prevent the crime. More laws aren't going to help the situation.

Re:Moderators, mod++ parent please!

[eean]

The irony is driping.

Moderators: mod parent++funny!!!1one LOL. ROTFL.

Or not, who cares.

presiding-preceding (n/t)

[eean]

presiding->preceding

Re:It's a "thin client"; it replaces nothing.

[ClosedSource]

So your management is conservative when it comes to industry standard PC clones that hundreds of companies have produced successfully, but they're liberal when it comes to non-standard desktop environments. Sounds ideal for your preferred solution, but I don't think most companies think like that.

Re:UNIX has had it even longer

[Dwonis]

IIRC:

GTK and Qt render almost everything on the client side, then push the resulting bitmaps to the server. They also do the same things to render text. Athena, on the other hand, tells the X server what to render, so the rendering is done on the server side, and huge bitmaps aren't being tossed around every time you click a button.

Re:UNIX has had it even longer

[Dwonis]

Um, nevermind my heresay. See X Window System Network Performance, by Keith Packard and Jim Gettys.

Re:It's a "thin client"; it replaces nothing.

[bluekanoodle]

No I think you misunderstand. To us, a quality pc is a pc from a reliable manufacturer for 799, not a cheap box for 350. Consider that a thin client appliance with licenses costs us less then 550 (with the same end user experience as a fat client). I'd rather spend our money somewhere.

800 for a corporate level pc is not uncommon. .

And don't tell me I could build one cheaper out of the same components as the big boys, because that's a lie once you factor in my time in assembling , installing and configuring the OS. Considering what I'm paying my staff, that's a huge waste of company resources.

If thin client is such a failure, and most companies don't think like that, then tell me why Places like Merril Lynch, t-mobile, chrysler, forbes etc use them? Oh that's right, because in your limited world view, you don't like them, so they must not make business sense

here's some more. http://www.citrix.com/site/aboutCitrix/caseStudies /caseStudies.asp/

Both client and server side are in since W2K

[melted]

That's how you do "remote desktop" and "remote assistance". If there was no server you wouldn't be able to do this. The missing part is the ability to connect, say a dozen users. For this to work you need to buy licenses.

I was just thinking about doing this...

[WoTG]

I just signed up for a VPS account a few days ago. It's FAR more power than I really need for my web hosting, so I was planning to setup a little Linux VNC login for myself. This would be great since most of the time I'm sitting in front of a Windows box, so I can get the best of both worlds. NX would be even better, though I doubt that I would be able to burn through my 50GB of traffic allotment with VNC...

Re:It's a "thin client"; it replaces nothing.

[Master of Transhuman]

I agree if we're talking about special-purpose hardware thin clients like the sort of thing Ellison's subsidiary sells.

Here we're talking commodity boxes which are slimmed down clients. Not the same thing.

Naziwin's law

[some guy I know]

There was a similar story about a Man whom people declared as being the Son of God

Naziwin's Law: "As a Usenet discussion grows longer, the probability of a comparison involving God or Jesus approaches one."

Re:Naziwin's law

[eean]

If influencing moderators means your aspiring to be God, suppose moderators are Gods themselves, walking amongst us. I'll remember that the next time I get mod points, maybe I'll cure world hunger or something.

Re:No, wrong idea

[kermit6306]

It's spelled "spelled". Not that I never make spelling mistakes; I'm just saying..

Sorry about that

[HangingChad]

Should have included the source, my bad.

From a Yankee Group study published on ZDNET:

zdnet.com.com/2100-1104_2-5330340.html

I like the way they say "only" 5% consider a total switch to Linux. I read 5% and thought HOLY SH--!! 5%!

Re:It's a "thin client"; it replaces nothing.

[ClosedSource]

Well, my point is that I believe the most companies feel there is less risk in buying cheap PC's than in adopting new OS's and desktop environments that their employees have to be trained in from vendors that may not be in business in a few years.

I'm not saying that all companies think like that, just that most do. You're not seriously suggesting that thin clients currently outnumber PC's are you?

Thin clients are probably popular in POS and other applications where repeatitive and limited functionality is being excercised. In these applications, employees only need to learn a limited set of activities and centralized processing is required anyway.

Re:Not all bible-thumpers are republicans

[GreyWolf3000]

Bible-thumpers are basically the rebirth of Calvinism within the US. (specifically the belief that good fortune is the sign of divine favor, and that ill fortune is the sign of damnation...

That part actually sonds more like Buddhism or Hinduism.

Re:Destroying NoMachine's business model

[agoliveira]

I'm a NoMachine developer and I can say that we (and this includes NoMachine's top managemment) praise and help the FreeNX initiative. Actually, they even have access to our internal mailing list where we discuss the latest plans.
We believe that FreeNX will help us in the future, not undermine the company.

may not be fixable

[jeif1k]

The problem is probably not fixable at the level of toolkits: their APIs weren't designed with that kind of usage in mind and KDE and Gnome application programs seem to assume local connections as well (e.g., in their redraw logic).

In fact, one of the reasons for the existence of the low-level parts of Gtk+ seems to be to pretend that the complexities that writing high performance client-server GUI apps entails don't exist (of course, pretending that doesn't make them go away).

What we really need is a server extension or tool that lets you simulate running applications on a slow, long-latency line. Then, application programmers can see the effects of their design choices.

Re:Not all bible-thumpers are republicans

[GreyWolf3000]

I disagree about Buddhism.

Modern American culture has a skewed view of it IMO--it is merely the rejection of the Vetas and of the traditional caste system in India. Reincarnation and karma are very much part of at least Theravata Buddhism (far mory widely practiced in the world), and from what I understand Mahayana Buddhism as well.

I think any criteria that does not classify Buddhism as a traditional religion--meaning that there are beliefs that are expected to be taken without question--would yield similar results for any other religion.

Calvinism makes more sense when you put it in it's historical perspective. The reigning variant of Christianity in Europe was Catholicism. The Counter-Reformation had just spread through Europe, and the whole movement to return Christianity to it's scriptural origins had largely backlashed.

In Calvinism, being born filthy rich means you're going to heaven pretty much no matter what. That's a major difference.

I'm not sure I think anyone believes that. Could you provide a source? What was all that about a camel passing through a needle? I was under the impression that early Calvinists tended to read the Bible much more than the Catholics of the day.

I ditched Orkut

[hummassa]

the FOAF spam killed it to me.

Re:Not all bible-thumpers are republicans

[Fished]

So when you see people complaining about "bible-thumpers", please substitute the word "Calvinists" instead of "Christians". You may find that the complaints have merit.

I think the word you're looking for is "Fundamentalists." Many of us who are not fundamentalist would maintain that we offer a much more Biblical worldview than the fundies manage.