Whatever happened to intelligence in the applicattion [Was: Re: The Qo s PipeDream]
I think you just tossed a red herring into the discussion. :-) I would suggest that a semi-intelligent playback bufferring scheme in the VoIP application, plus a 'semi-lossless' link, would be just fine. ;-) Doesn't anyone really remember the whole smart-v.-stupid network analogy? Not meaning to start a flame war here, but trying to stick all of the intelligence back into the network is not exactly a win-win proposal. - ferg -- Mikael Abrahamsson <swmike@swm.pp.se> wrote: On Fri, 16 Dec 2005, Christopher L. Morrow wrote: When you're running voip over a T1/E1, you really want to LLQ the VOIP packets because VOIP doesn't like delay (not so much a problem) nor jitter (big problem), nor packetloss (not so much a problem if it's less than a 0.1 percent or so). So combining voip and data traffic on a link that sometimes (more often now when windows machine have a decent TCP window) go full, even just in a fraction of a second, means you either go QoS or do what Skype does, crank up the jitter buffer when there is high-jitter, which means latency for the call goes up. [snip] -- "Fergie", a.k.a. Paul Ferguson Engineering Architecture for the Internet fergdawg@netzero.net or fergdawg@sbcglobal.net ferg's tech blog: http://fergdawg.blogspot.com/
On Fri, 16 Dec 2005, Fergie wrote:
Doesn't anyone really remember the whole smart-v.-stupid network analogy? Not meaning to start a flame war here, but trying to stick all of the intelligence back into the network is not exactly a win-win proposal.
Yes, in a perfect world you're correct, but the equipment which does POTS<->VoIP is probably easier (and cheaper) to do if it has a static 40ms jitterbuffer instead of having to constantly adapt. Echo-cancellation probably is easier to do if delay is constant (which it is with constant jitter-buffer). OTOH the echo cancellation should be done in the POTS-VoIP interconnect and the echo is only present in the POTS part so that should work anyway. But then again, that's why Skype is to successful, it adapts to network conditions, so it might be that this way will prove more successful. -- Mikael Abrahamsson email: swmike@swm.pp.se
On Fri, 16 Dec 2005, Fergie wrote:
Doesn't anyone really remember the whole smart-v.-stupid network analogy? Not meaning to start a flame war here, but trying to stick all of the intelligence back into the network is not exactly a win-win proposal.
Trying to stick it all in the application is not exactly a win-win proposal either. The problem with religious dogma is it leads to a lot of burning people at the stake, for more stupid analogies. Finding the right blend of what applications can do well, what the network can do well, and how they interact is the challange. For example, smart applications don't handle DDOS attacks very well, and regardless of how much network capacity you provision, there is always a DDOS that is bigger.
Thus spake "Fergie" <fergdawg@netzero.net>
I think you just tossed a red herring into the discussion. :-)
I would suggest that a semi-intelligent playback bufferring scheme in the VoIP application, plus a 'semi-lossless' link, would be just fine. ;-)
Any competent VoIP application/device developer will use an adaptive jitter buffer. It's really not that tough, and most apps/devices have them today because working products sell better than non-working ones. My VoIP phone (full disclosure: I work for the vendor) operates just fine at home over a DSL line, across four ISPs, through two NATs, and to a gateway in Canada. The voice gets a little choppy when a 10MB powerpoint hits my Inbox (sadly, several times per day), but it self-corrects after a couple seconds.
Doesn't anyone really remember the whole smart-v.-stupid network analogy? Not meaning to start a flame war here, but trying to stick all of the intelligence back into the network is not exactly a win-win proposal.
I think you'll get further by arguing that intelligent networks with small pipes cost more to maintain than dumb networks with fat pipes. Less likely to induce sleep in your bean-counters. S Stephen Sprunk "God does not play dice." --Albert Einstein CCIE #3723 "God is an inveterate gambler, and He throws the K5SSS dice at every possible opportunity." --Stephen Hawking
On Friday 16 December 2005 09:21, Fergie wrote:
Doesn't anyone really remember the whole smart-v.-stupid network analogy? Not meaning to start a flame war here, but trying to stick all of the intelligence back into the network is not exactly a win-win proposal.
A stupid network is easier for malicious applications to exploit. Need a balance point, not either extreme. -- Lamar Owen Director of Information Technology Pisgah Astronomical Research Institute 1 PARI Drive Rosman, NC 28772 (828)862-5554 www.pari.edu
participants (5)
-
Fergie -
Lamar Owen -
Mikael Abrahamsson -
Sean Donelan -
Stephen Sprunk