[e2e] Is a non-TCP solution dead?
Cannara
cannara at attglobal.net
Wed Apr 2 09:13:06 PST 2003
Ian, I'm afraid we seem to get too anxious about fitting data to our
experience. Loss is not always due to 'congestion' -- which, in fact, has
many forms. Congestion is not alyways a cause of loss. The .4% I mentioned
was simply a flakey physical interface on a 2xT1 leased telco link to a Cisco
router -- silicon fails now and then.
Alex
Ian Wakeman wrote:
>
> >>>>> "cannara" == cannara <cannara at attglobal.net> writes:
>
> cannara> Now, add in a 0.4% loss on the path,
>
> But I thought you were asserting that the network was run by highly
> efficient telco switches and lines, which wouldn't drop traffic.
> Surely this 0.4% loss wouldn't be coming from (gulp) congestion in the
> network? Which may have been made the loss rate worse by running the
> flows at higher rates, and more importantly, made the loss rate higher
> for everyone else using the bottleneck link?
>
> cheers
> ian
--
Alex
79xj6L SII (BRG + wires)
86xj6 SIII (Black)
61 Sprite MkII (Red)
Menlo Park, Calif.
More information about the end2end-interest
mailing list