On 9/18/07, Jan Hoevers <[EMAIL PROTECTED]> wrote: > sjm wrote on 18-9-2007 15:27: > > Is this not true only for the case where one is saturating the link and, > > therefore, being throttled to the link max speed. If one where only > > using 100kbits up/down in the above example, would things not be > > symmetrical and only start to be asymmetrical when the traffic reached a > > point where one side is being throttled to the speed? > > > > Just wondering.
These are two separate situations. What I was discussing in my previous email was the case where there is no other traffic at all on the line. Even when this is the case, since DSL lines (like most types of data access lines) only transmit one bit at a time, it takes a certain minimum amount of time to transmit a packet of a certain size, and this minimum is determined by the bitrate of the line. Since the upstream and downstream bitrates differ (with the upstream bitrate being slower in almost all cases), it will take more time to transmit a packet of a given size then to receive one. Queueing of packets in a router or modem (bridge) when a link is saturated is a different (and, for a DSL line, usually much larger) source of latency than serialization delay. As E Frank Ball has mentioned, a setup like the Wondershaper can mitigate the problems caused by a congested DSL upstream to a large extent. When I was using a DSL line, I ran a Wondershaper-like script, but only on the upstream since that was the only direction that was ever noticeably congested on my line. Rusty _______________________________________________ timekeepers mailing list [email protected] https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers
