Hello Kurt, [snip] > My problem is the downstream, e.g. a HTTP download. I thought it should > be possible to lower the incoming datarate by using a smaller TCP > window, resulting in the remote host waiting for acknowledgement of each > packet and therefore reducing the datarate. Maybe together with delaying > the TCP-ACK packets if there are a lot of downloads. apart from other doubts expressed here on the list this also wont work when someone is downloading UDP based traffic... > > I did already try the traffic shaper, keeping my upstream bandwidth > below the line bandwith thus having empty upstream queues all the time. > But it did not fix the jitter especially on downloads, so I think > something else must be done as boxes like the AVM Fritz VoIP box is > quite unaffected when up-/downloading while using VoIP. are the TOS/DiffServ fields of the incoming stream set to some useful values? This may help the ISPs router to give the Voip packets better priority. Also the jitter-buffer algorithm of the AVM box may just be better. Assuming that some priorization is working a 1500 byte packet should cause something around 20 ms additional delay in worst case - which isnt that bad... You may test the different buffer algorithms with tools like http://snad.ncsl.nist.gov/itg/nistnet/ or http://www.wanulator.de in your lab. Good luck and best regards, Michael - : send the line "unsubscribe linux-net" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
