[e2e] Lost Layer?

Detlef Bosau detlef.bosau at web.de
Sat Jan 11 04:50:42 PST 2014

Am 11.01.2014 12:40, schrieb Jon Crowcroft:
> In missive <52D04B36.9010005 at web.de>, Detlef Bosau typed:
>  >>I would like to discuss the talk
>  >>http://rina.tssg.org/docs/JohnDay-LostLayer120306.pdf
>  >>given by John Day.
>  >>
>  >>What do you think, e.g., of the claim
>  >>> •
>  >>> TCP was split in the Wrong Direction!
>  >>> • It is one layer, not two.
> should have been 3 - as per the transport services work - its clear

Why 3? Why not 4? Or 40?

Admittedly, sometimes I think CS guys are collecting layers just like
others are collecting stamps.

> you need a sublayer convergence (as per day's work) but also the
> socket layer needs revising badly to allow for a wider set of
> transport service semantics than came out of the fast 
> hack that bbn and berkeley did

Shouldn't we agree upon a model and then upon offered services and APIs

When I read John's talk, my impression was (and this is not only from
John's talk), our current <small>TCP</small><big>IP</big> model is
roughly spoken as follows.

<tiny>TCP peer</tiny><HUGE BOLD LARGE>I P C L O U D</HUGE /BOLD
/LARGE><tiny>TCP peer</tiny>

And our current end to end excuse is: We only can control the TCP peers,
there is a big fat IP cloud where we have absolutely no idea what's
happening in there, so believe in a model instead and control this at
the end points.

In other words: We mix up the real IP cloud with an artificial model -
and ignore that these are not really the same.

And perhaps, that's what John wanted to say? (Which is sometimes a bit
ardous, because John often talks extremely much about the Internet's
history and where this company took the wrong road or when this
institution took a wrong decision, this may be interesting and sometimes
worthwhile to understand how things developed, however it is sometime
not quite concise and we sometimes don't really see the forest for the
trees here.)

Actually, the more I think about it, I have a huge problem with this "IP
cloud", because this metaphor withholds what's going on there and the
end points are, and this is simple control theory, NO WAY adequate to
control the mechanisms within a cloud of such complexity.

Detlef Bosau
Galileistraße 30   
70565 Stuttgart                            Tel.:   +49 711 5208031
                                           mobile: +49 172 6819937
                                           skype:     detlef.bosau
                                           ICQ:          566129673
detlef.bosau at web.de                     http://www.detlef-bosau.de

More information about the end2end-interest mailing list