Re: [v6tc] Issue 1: Could we live with UDP encapsulation always on?

[Jeroen Massar <[e-mail]>]

On Fri, 2004-12-17 at 03:20 -0800, Alain Durand wrote:
> On Dec 17, 2004, at 3:11 AM, Jeroen Massar wrote:
> > You mean eg dailup users? Do those still exist? Also G3 phones don't
> > have this issue either with UMTS and other goodies being deployed,
> > coming upto speeds of most lower DSL lines...
> 
> The issue is not so much speed but money. If there is a 7-10% overhead,
> this is not insignificant, regardless of bandwidth.

In general for 'office/home environments' aka webbrowsing and alike, it
seems the trend is that packet length distribution is that about 40% of
the packets is between the 0-400 size. the other 60% is in between
400-1200 with only relatively few packetlengths in tehe 1200-1400 range,
one will have a peak for the 1500 size, thus completely full packets.
On average, from the couple of graphs that I have I would estimate the
packetlength to be around 600 bytes/packet.

This is based on an ethernet (mtu=1500) environment though, thus the
peek at 1500 will move downwards, relative to the mtu size of the
environment. Also, when there is a high concentration of 'downloaders'
it might well be that that the packetsize distribution shifts almost
completely to the maximum mtu. All depends on the usage of the network
and if you count packets or octets as an importance.

There will always be some sort of overhead with encapsulation. But if
you are thinking of that, make a new protocol number which would make it
proto-41 or GRE, which did not work over NAT's.

> Please keep this discussion for the upcoming issue 3 thread:
> "Should some of this authentication mechanism "follow-on" after the 
> tunnel set-up phase as finished?"
> To keep the discussion focussed, I will only introduce one issue at a 
> time.

Ack, though if you want to minimize the header, you can't introduce any
of those parts anymore because the header is to small ;)

Greets,
 Jeroen

_______________________________________________
v6tc mailing list
[e-mail]
https://www1.ietf.org/mailman/listinfo/v6tc