Slashdot Mirror
Intel Whitepaper On UPnP
An anonymous reader writes "This article by two developers at Intel provides an introduction and overview to Universal Plug and Play (UPnP), a standards-based technology for transparent network device connectivity that allows devices from various vendors to "just work" when plugged into the network, eliminating the administrative hassle typically associated with networking devices and making them programmable entities that can be controlled across the network. Intel has been a strong supporter of UPnP, and has released an open-source SDK for the development of Linux-based UPnP devices, hosted at SourceForge, which has been used in a number of UPnP products that will soon show up on the market."
Has anyone posted a detailed comparison of UPnP and Zeroconf/Rendevous? They seem to accomplish similar things.
Please reply to this comment with any links.
These systems (Rendezvous, UPnP, etc.) seem to neglect one issue that's important to me personally.
If I plug something in to my network, I want to know exactly what it's doing and what it's not. Unless I tell it otherwise I want it to sit there and do absolutely nothing. Am I missing something here? The last thing I can imagine being useful is for "intelligent" devices to start making decisions about what they think I want them to do.
Meep meep
Good one on the part of Intel with comming out with a SDK for linux as well for other platforms. This is exactly the sort of stuff we need to see more often from major companies. It helps to keep linux up to date from the people who make this stuff, not having the end users rely on people to hack apart the specs themselves and write drivers from scratch. Good one Intel.
What about using atleast a standard ?
SLP - service location protocol. RFC 2608
Take a look at e.g. www.openslp.org and www.srvloc.org
Once OpenSLP implements RFC 3011 , cool things can be done,
e.g. get rid of stupid ipv6 dhcp servers, and serv configuration through SLP.
Intel has done two things wrong with this: 1) Reinventing the wheel. Zeroconf is already there, and open-source too. You'd think that with a "universal" standard, you'd at least want to go with what's already there. 2) Trying to shove on Apple again. This time, it's going to be harder though, because Zeroconf already is supported on many devices and applications, including the TiVo. Another reason to just use Zeroconf.
Does anybody have dates corresponding to various milestones of UPnP and Zeroconf. I heard of UPnP before I heard of Zeroconf. However, it's gone a long time without many products that I have seen.
SLP is deprecated in favor of Rendezvous.
Glad to see you have survived at Intel, and are having fun toys to play with still... I prefer faster technologies than generally make it home
I have mod points and I am not afraid to use them
Considering the source: M$FT/intel... how likely is it to be a patent/royalty-free "open" standard? Who's on the technical committee? I love it when win/tel secretly develop a standard in a black-box environment, then get ISO/IEEE to rubber-stamp it. You get good things like DDE/DDX, OLE, OLE2, ATL, COM, ActiveX, and VBS. Well, I guess Sun is guilty for that too, w/ Java.
"All your desktop are belong to Gate$."
The biggest trick the devil pulled was letting lawyers become politicians so they can write the laws.
My better request to AMD: many independent buses for HDs, Gigaethernets, TV, SVGA and FPGAs with the condition of good "Z impedance".
JCPM (c) (kloneright)
Quoting the article:
Unsecured devices continually volunteering personal information to the outside world? I don't want:
On second thought, I take all of that back. Come to think of it I have decided I like UPnP. Not only can I viewz0r my neighbor's electronic picture frames, I can give them, and their noisy kids, something really educational to look at.
...An AMD ad, of course. Take your mind out of the gutter!
UPnP: providing remote administration to Windows XP since 2001.
This is a pretty "new" technology. There is not that many resources/ user group about it. ...?
I have a Microsoft wireless router (MN-500) and it supports UPNP.
I started looking into it and it looks like there are a few cool COM objects that can be used in conjunctions with Advanced XML namespaces to do intersting things programatically.
I wonder what kind of security they put into it
One of the feature/use of UPNP is to traverse and handle NAT properly.
The thing that really bug me, is that they keep adding MMSG ports ( I guess they are from MSN messenger) to my static list of redirected ports.
Geez, stick with one, don't take 10/20 of my port redirections.
All I need is for one of those ports to become easily exploitable and I'm toast !
I can see why it's nice:
MY Webcam/ VOIP/ Remote Assistance, all of the features of MSN Messenger on XP (some avail on other versions) just work.
I still would like to keep it to a few static, non changing ports.
Anybody else experienced that kind of behavior ?
On http://www.microsoft.com/WINDOWSXP/pro/techinfo/pl anning/networking/nattraversal.asp there is a rather interesting article explaining the basics & problems associated with NAT. For beginners. At the end they explain the Windows API for UPnP.
The list with Limitations of NAT Traversal is funny, were it not that this will open a cornucopia of new virus possibilities. NAT Traversal is one of the possible solutions in a UPnP device:
NAT Traversal technology has been created to enable network applications to detect the presence of a local NAT device. Once detected, the application can then configure the NAT, defining the appropriate mappings to solve their compatibility issues.
So far so good. However:
* NAT traversal has an open trust model. This means that all application on the private network have access to all the port mappings on a NAT. This allows for a great amount of flexibility of multiple points of administration, but applications do not have exclusive ownership of their mappings.
Oh goody, Open Trust, sounds like an MSFT technology enhancement!
* Conflict resolution is the responsibility of applications. If an application tries to map a port that is already mapped to another client, it is up to the application to either find another port or overwrite the application.
This means that if a rogue email client wants the smtp port redirected to its office machine, the real email server doesn't get any mail, because it is the responsibility of the rogue application!
* Applications are responsible for cleaning up after themselves when they are done with a port mapping. Static mappings persist indefinitely and are most appropriately used by services that intend to listen on well-known ports for the life of the application.
This means that on top of memory leaks, we can have portmap leaks: applications that are not written correctly will 'hog' the UPnP router ports, spoiling it for others.
I first encountered this UPnP thingy when a colleague of mine couldn't use video chat with his MSN Messenger client. It seems MSN wants to have UDP portmaps from, 5004 - 65535 mapped to the PC it is using. Funny that. Most other video clients only use 2 ports, at most. Instead of improving their MSN protocol to play nice with other network clients, MSFT invents a new, Universal Plug & Play standard!
I think I'll stick with iChat.
...a hack!!!
Microsoft/Intel are HOLDING THE WORLD BACK by in introducing somthing that will make that sickening hack called NAT hang around like the bad smell it is.
NAT is like a big smelly shit in a toilet. This UPnP crap is like opening the door to the toilet and smelling the fart smells. IPv6 is like flushing the toilet and spraying fragrant toilet spray.
Don't introduce UPnPray, (pray for securities sake), introduce IPv6 GOD DAMN YOU INTEL and MICROSHAFT!!!