Path: utzoo!censor!geac!torsqnt!news-server.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!cs.utexas.edu!samsung!uakari.primate.wisc.edu!ames!excelan!donp From: donp@na.excelan.com (don provan) Newsgroups: comp.protocols.tcp-ip.ibmpc Subject: Re: ARCNET <--> Ethernet Message-ID: <1087@excelan.COM> Date: 23 Mar 90 18:17:07 GMT References: <9003160914.aa08039@louie.udel.edu> Sender: news@excelan.COM Reply-To: donp@novell.com (don provan) Organization: Novell, Inc., San Jose, Califonia Lines: 19 > 3. I have seen reports of people trying to implement TCP/IP on top of >ARCNET. What seems to be the universal problem is the size of the MAX packet. >ARCNETs limits require breaking the packets and this seems to be a problem. > > That is about all I have found to date. Anyone else care to comment? While ARCNET itself has a limit of 504 bytes of data per frame, there is a fragmentation/reassembly standard for exchanging packets up to 60,480 bytes in length. (There's also an older standard which allows packets with 1008 bytes of data carried in two fragments. The older standard is a subset of the newer standard.) Unfortunately, neither scheme is compatible with the ARCNET packet format described in RFC1051. The longer packets make ARCNET quite usable as a vehicle for IP traffic. We're planning on writing an RFC to replace RFC1051, reflecting these fragmentation techniques. If you're interested in ARCNET, feel free to send me input. don provan donp@novell.com