Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!tut.cis.ohio-state.edu!ucbvax!proteon.com!jas
From: jas@proteon.com ("John A. Shriver")
Newsgroups: comp.protocols.tcp-ip.ibmpc
Subject: SNA over Ethernet via NDIS in OS/2 EE environment
Message-ID: <8905302112.AA01352@monk.proteon.com>
Date: 30 May 89 21:12:33 GMT
References: <8905301958.AA27545@saturn.acc.com>
Sender: daemon@ucbvax.BERKELEY.EDU
Organization: The Internet
Lines: 23

When you run SNA over Ethernet (or IBM Token-Ring, a/k/a 802.5), you
use 802.2 class 2 LLC.  Virtual circuits.  They plug in where SLDC
fits in the normal SNA stack.  Uses SAP's 04, 08, and 0C.

Address translation?  Haven't you ever seen somebody do an NCP gen?
You have to configure all the MAC addresses into all the routing nodes
in the SNA network, along with all the possible routes, and their
costs.  (That's right, if someone replaces an Ethernet or Token-Ring
board in a PC, the network manager has to do NCP gens.)  There is
nothing dynamic in present-day SNA.  Maybe in the future LEN version
of SNA.

But that's OK, all of IBM protocol stacks for the PC use 802.2 class
2.  NETBIOS and SNA both do.  Why do you think they are stuck doing
bridging?  An SNA router is rather pricey (a 3725 is two bays, and
easily over $100K.)  Thus, their NDIS driver will probably include
access to 802.2 class 2.  Of course, if it's not in board microcode
like Token-Ring has, it will run a lot slower on Ethernet.

Of course, the only "mainframe" that speaks SNA over Ethernet is still
the 9370.  (What purpose is there for SNA except to speak to a
mainframe?)  All of the others only speak SNA over Token-Ring (and
SLDC) via 3174 or 3725/3745.