> > Thx for the response. I've considered that but ruled it out. Doing a > packet capture on the HT end, I am sending about 30-36 pkts per second, > totaling about 4500 bytes on the wire per second. So I am only transmitting > about 5 Kbs. Even a low upload speed could handle that. >
Maybe it's caused by something else. However, I believe that 5Kbs could cause speed adjustment in TCP over some ADSL connection. It's enough to have a very little more delay between 2 RTP packet in order to create an increasing latency over the time. > This is more than an 'inconvenience' for using TCP for voice traffic. It > would make TCP unusable for voice traffic. I have worked with other > products that did SIP/RTP using TCP and we did not have this delay problem. > I myself had this problem from times to times but generally my communication with http tunnel worked well. > > > Jeff > > ----- Original Message ----- > From: "Minh Phan" <[email protected]> > To: "Jeff Theinert" <[email protected]> > Cc: [email protected] > Sent: Tuesday, July 7, 2009 1:02:49 AM GMT -08:00 US/Canada Pacific > Subject: Re: Http Tunnelling > > Hi Jeff, > > Is your local end using an ADSL connection? If it is the case, the > increasing latency might be caused by the limited upload speed. > If the connection speed from your phone to the HTTP server is low, the TCP > protocol would naturally increase the delay between packets in order to > avoid packet loss. Hence, increase the latency. This is the inconvenience of > using TCP for voice traffic. > > Best regards, > > Minh > > > > 2009/7/7 Jeff Theinert <[email protected]> > >> Hi, >> >> >> >> I have HTTP tunnelling working under Windows, but there is a problem I >> cannot resolve. During a call where the local end is using HTTP >> tunnelling, the audio latency to the remote end degrades by about 1.6 >> seconds per minute the call lasts. The audio latency from remote end to >> local end is consistent. I checked the server code and found that it was >> compiled with debug messages, but even recompiling to eliminate the debug >> messages did not change the degrading latency. >> >> >> >> I cannot find anything specifically causing a delay in the client. If I >> set a breakpoint in Visual Studio and let the program sit for 20-30 seconds, >> when I resume, the latency is back to normal, but then begins to degrade >> again. Since all threads in VS are frozen at that point, that makes me >> believe this is a server issue. But since the server code has not changed >> in 3 years, I would expect that some one would have reported this as a >> problem. >> >> >> >> Any ideas to resolve this are greatly appreciated. >> >> >> >> Jeff Theinert >> >> _______________________________________________ >> QuteCom-dev mailing list >> [email protected] >> http://lists.qutecom.org/mailman/listinfo/qutecom-dev >> >> >
_______________________________________________ QuteCom-dev mailing list [email protected] http://lists.qutecom.org/mailman/listinfo/qutecom-dev
