uucp@tmsoft.uucp (Now THATs a Daemon) (03/22/89)
# >Topic: I've been confused by much of what I've been reading about
# >'migration from TCP/IP to ISO'.
# >
# >Is this necessary?
#
# YES!!! I think it will make things simpler. Believe me! :-)
#
Personally, I doubt things will be simpler. Working with networks as
large as the ones at UofT has changed some of my perspectives.
It's my feeling that ISO networking will come into popular use after
Berkeley releases 4.4BSD. The NSB (now called something else but I
forget what) is funding this effort, and the GOSIP grace period should
begin the day 4.4BSD is released.
ISO networking at level 3 is not yet useful. Heck, they haven't
even decided on a IGP yet.
# If you want ISO now, and backwards compatability, and transparent
# upgrages, or just plain good neworking, you should be looking at
# SysVr{3,4}. I don't see OSF implementing STREAMS or anything like it,
# and as such I doubt their networking implementations will be easy to
# understand, easy to change, or easy to implement in the first place.
Umm, System V release 3? Not even SunOS 4.0.X with Van Jacobsen's mods
is just plain good networking. Most things work right, but there are still
a few problems. I haven't seen SysVr4, but I have my reservations that
it can be used as a decent router. Perhaps I misunderstand what you mean
by networking. I don't regard BNU or RFS as such.
Don't be quick too quick to judge AIX. The IBM researchers in Milford
and Yorktown are at the leading edge of WAN network technology -- IBM is
very fortunate to be involved with the NSFnet to the degree that they
are. The programmable routers on the NSFnet use 4.3BSD tahoe, but they
will migrate to AIX. I would not be surprised to see good code in AIX
after that.
To date, the ISO network applications I've seen are monsters. The file
transfer utility sources alone may be larger than my kernel sources.
The application gateways will be something else as well. I'm in no
hurry thanks.
Tom Molnar
UofToronto Computing Services Unix Systems Group