Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!rutgers!cs.utexas.edu!wasatch!haas From: haas@wasatch.utah.edu (Walt Haas) Newsgroups: comp.dcom.lans Subject: Re: NOVELL networks Message-ID: <2004@wasatch.utah.edu> Date: 8 Jun 89 18:05:21 GMT References: <11806@s.ms.uky.edu> <161@uvacs.cs.Virginia.EDU> <9585@j.cc.purdue.edu> Distribution: usa Organization: University of Utah, College of Engineering Lines: 26 In article <9585@j.cc.purdue.edu> tim@j.cc.purdue.edu (Timothy Lange) writes: > ... I do know that Novell Ethernet networks out of the box are 802.3 specs. > But they supply a program that changes the packets so they will use their > Xerox assigned packet id of 8137. Does this make sense? I haven't been > able to find someone that can tell me the difference between 802.3 and > others. The difference between Ethernet packet format and IEEE 802.3 packet format is that: * in an Ethernet packet the first 16 bits after the addresses contains the protocol type, and you have to infer the length of data. * in an 802.3 packet the first 16 bits after the addresses contains the length of the data, and you have to have other means to determine the protocol. On a physical Ethernet that carries packets of various protocols, e.g. Novell, DECnet, IP etc., the receiver of a packet must be able to determine the protocol. This is done by testing the protocol type in Ethernet packets. In 802.3 packets the problem is a little harder -- it is customary for the 802.3 data field to in turn contain an 802.2 packet whose header specifies the protocol. For example IP would be carried in an 802.2 Type 1 LLC datagram with SSAP and DSAP values indicating IP. Unfortunately Novell apparently didn't use this convention, making it impossible to use their 802.3 framing on a network with other 802.3-framed protocols. Therefore it's safest to configure your Novell network to use Ethernet framing. Cheers -- Walt Haas haas@cs.utah.edu utah-cs!haas