From: Jon Crowcroft <J(_dot_)Crowcroft(_at_)cs(_dot_)ucl(_dot_)ac(_dot_)uk>
...
i also like the fact that MOST fancy things we do in traffic treatment
(firewall, diff/int serv, header compression, checksum) ignore this
layering rubbish idea and look at the whole packet and the whole state
of the systems where you need to......creating articifical monitor
security/concurrency boundary crossings in component software was a
terrible thing, and many modern OSs have got a long way away from
it....
on reflection (pun intended):
layering is a clothing methodology for people who live in fractal climates
not a design pattern for correct or efficient communications
systems....
Yes, but without the ISO layering terminology, what would we have done
with all of the time wasted arguing with layering priests about what
is impossible or sinful? Without the layer labels, how would legions
of the pointy haired, salescritters, and people who implement TCP by
installing software on PCs prove their expertise?
Without the layering religion, much of the steam in the GOSIPs might
not have been generated. Do you remmeber the endless educational lectures
from experts informing us that the OSI suite was inevitiable and desirable,
mostly because it had proper layering unlike that ad hoc DDN stuff?
A serious case can be made that the 7 layer model did more harm than
good by helping the congenitally or willfully clue challenged pretend
to be knowlegable and authoritative. Without the simplistic view of
networking required by the 7 layer model, the experts might have been
forced to find other areas to plow.
It's not that the good in layering needed to be invented by the ISO
and then imported into the rest of the network universe. The notion
of software and hardware modularity had already had its 15 minutes of
being the panacea among the CASE gurus by 1976. (Yes, I know that
species of psuedo-expert didn't call itself that then and doesn't
today.)
Vernon Schryver vjs(_at_)rhyolite(_dot_)com