NAP/ISP Saturation WAS: Re: Exchanges that matter...

alex at alex at
Wed Dec 18 09:45:45 UTC 1996

As for me, congestion exists if long TCP sessions can't adapt their
window sizes and packet drop if more than 3 - 5 % over connection line.

This means more than 1,000 - 2,000 tcp sessions through 2 Mbit link, for
example. Yes, IP is connectionless protocol, but Internet is not connectioonless
network in real life...

>   Craig Nordin  <cnordin at> wrote:
>   >> Er.  There's no such thing as perfect transport as long as TCP is
>   >> concerned.  If end-hosts support large windows, even a single
>   >> TCP session will load the network to the point where it'll lose
>   >> packets.
>   >What you say here makes sense to me.  But, out of a 500
>   >ping sample over the course of half a day, I was getting well
>   >below one percent loss -- six months ago.
>   Most TCP sessins are pretty short, in fact.  They simply have
>   no chance to open window.   Also, there's a lot of old TCP
>   implementations which have window limited to 64Kb (or less!).
>   >Now, my same sample group is going for more than three percent,
>   >with many ten percent loss routes showing...
>   Well, nobody argues that Internet is overloaded.  But the question
>   of how to measure the congestion is not as simple as just pinging.
>   The problem is that there's no reasonable way to learn loss
>   statistics from routers.
>   --vadim

Aleksei Roudnev, Network Operations Center, Relcom, Moscow
(+7 095) 194-19-95 (Network Operations Center Hot Line),(+7 095) 239-10-10, N 13729 (pager)
(+7 095) 196-72-12 (Support), (+7 095) 194-33-28 (Fax)

More information about the NANOG mailing list