Re: Working with TR-64
Re: Working with TR-64
- Subject: Re: Working with TR-64
- From: Josh Graessley <email@hidden>
- Date: Thu, 23 Mar 2006 11:14:51 -0800
UPnP is being used to do everything under the sun. One aspect is
creating NAT mappings. UPnP is an unwieldily beast. Having to support
UPnP in a base station just to implement NAT mapping is a large
burden. On top of the issues with including a buggy XML parser (is
there any other kind?) and all the other crap you need for UPnP, the
way the protocol is defined for creating port mappings is seriously
flawed.
The UPnP way of creating a port mapping is to use a query to find out
if the port you want is taken. If it isn't, you perform another
operation to take that port. There is a synchronization problem here.
If, between checking for the port and requesting the mapping, someone
else does the same thing, you will overwrite a mapping. This doesn't
sound like it's a huge problem, what are the chances of that
happening, right? Well, try plugging in a switch that got unplugged.
Intelligent devices that recognize the link has come up will go
through all of this pretty quickly at the same time. If I remember
correctly, there were other problems. You could modify other devices
port mappings. There were also no timeouts on port mappings.
These are some serious flaws that can lead to real world problems and
tech support calls. As much as I love using established standards, if
they don't really solve a problem or solve it well enough, I can see
the need roll a better solution. I would certainly encourage others
to use NAT-PMP. Is it likely that a lot of people will? Maybe, maybe
not. On the other hand, when someone buys Apple's products and uses
them together, they will operate better than the competition. The
products will just work. mDNSResponder will fall back to using UPnP
for port mappings if that's all that is available. The network
connectivity through NAT experience will then be just as flaky as any
other UPnP device.
As for the services provided by zero config networking versus what
UPnP provides, there's a pretty good summary here:
http://www.zeroconf.org/ZeroconfAndUPnP.html
-josh
On Mar 23, 2006, at 7:46 AM, Marc Epard wrote:
on 3/23/06 9:31 AM, Peter Sichel wrote:
IETF zero config networking (Rendezvous/Bonjour) provides similar
services, so I'm not sure what advantage TR-064 offers.
TR-064 is a standard implemented by most if not all DSL hardware
vendors.
UPnP is a standard implemented by most small router vendors. Both
pre-date
NAT-PMP and I'm not sure if any devices other than Apple's base
stations
support NAT-PMP. I'm not at all fond of UPnP or TR-064 and I
really like
the simplicity of Apple's proposed standard, but this is likely a
case where
Apple should give up and support the popular standards.
-Marc
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Macnetworkprog mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
40apple.com
This email sent to email@hidden
Attachment:
smime.p7s
Description: S/MIME cryptographic signature
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Macnetworkprog mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden