Xref: utzoo comp.sys.att:3706 comp.sys.ibm.pc:16971 comp.dcom.lans:1544 Path: utzoo!utgpu!water!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!mailrus!ames!ucsd!sdcsvax!ucsdhub!jack!nusdhub!rwhite From: rwhite@nusdhub.UUCP (Robert C. White Jr.) Newsgroups: comp.sys.att,comp.sys.ibm.pc,comp.dcom.lans Subject: Re: Does anybody know anything about PMX/Term from AT&T? Message-ID: <1094@nusdhub.UUCP> Date: 6 Jul 88 22:26:13 GMT References: <740@cgh.UUCP> Organization: National University, San Diego Lines: 125 HI! It's time to play that wounderful game, sleuth or consequences! The unpleasant issue of the day, STARLAN OSI protocols! (the information in revealed in this is my "best research" on the topic, and was done on myown networks behalf. You will probably not like the answers.) In basic summary: all of the "old" STARLAN stuff will _not_ talk with any of the new stuff; the STARLAN and the STARLAN 10 are _not_ wire compatable, but any kind of bridge will make this relationship hunky-dorry; there is _no_ hardware difference between the old and new STARLAN boards, but the STARLAN 10 stuff is an entriely different game (it should therefore be possible to replace the drivers for non-suported systems). THINGS YOU WILL LOOSE: ISN SLIM-C cards (to be eventually replaced) RS232-C NAU (totally history, no future plans to replace _ever_) 3B1 Connections (by omission, no word as to the future?) STARLAN 10 _will not_ dasy-chain. NRUs are no longer necesssary for long runs. (NHU now does this function.) THINGS YOU WILL GAIN: NHU (Network Hub Unit) differs in important diagnostic and protection functions from the NEU which is still useable for STARLAN. The major feature of the NHU is that it will isolate faulty network sub-segments, increasing system integrity. 6386 Servers can "host" printing for any printer attached to any client. This will not be available for 3B server software. STARLAN and STARLAN 10 may be intermixed with ETHERNETdevices, or bridged to such segments. This does not include TCP/IP and others, but it dosn't seem to preclude it either new STARLAN drivers are (aledgedly) about twice as fast as the older stuff. All Wire-feet distances are increased. X.25 Bridging and SNA-STARLAN session bridging. Servers can generate alert messages on (MS-DOS) client screens. (i.e. warnings, alerts, and system condition messages) (now for the questions...) in article <740@cgh.UUCP>, paul@cgh.UUCP (Paul Homchick) says: > AT&T equipment in it, and there is a 6386WGS on my desk awaiting > integration. If there is a helpful AT&T Network Guru out there I have > a few questions: > > 1. Can the "PC6300 Network Program, Ver 2" communicate with one of the > Ver 3 Servers? NO. The "new" packet structures are totally different than the older structures on the most primitive level. while it should be possible to make a packet-type-translating bridge, there is no aparent intent to do so. > 2. Does the DOS client software come bundled with the Ver 3 Server > programs? YES. The "dos server program, Version 3" is similar to the network software you are used to receiving, but it is now sold with a minimum client licence of 8. (i.e. you pay more cash.) This seems to be a response to people liberally copying the client software. I do not know if there is a built in protection against ilicit copying(??) > 3. Can the 'old' STARLAN protocol and the new version 3 exist on the > same physical network? i.e: can there be a set of version 3 servers > and clients, and version 1 servers and clients on the same network? NO. The packets and such from the "old" and "new" versions are supposed to be capible of totally scrambling eachother. I have not tried this, so it may not actually be true. (This is supposed to be an addressing issue.) > 4. Will a 1:10 bridge provide connectivity between version 3 and version > 1 as well as connecting STARLAN and STARLAN10? NO. The 1:10 bridge preforms promiscuous address evaluation on all the packets, and then retransmits necessary packets on the far side of the bridge compleetly unchanged. Only necessary traffic is passed, so old format traffic would be filtered by default. An other bridging no-no is connecting lan segments in a circle. (i.e. A || B || C || A ) If there is more than one way to get there the birdges will storm. > 5. Is the "OSI Network Program" required to run the "Version 3 Server > Program"? If so, is it bundled with the Server, or is it an extra > cost item? .NA. The Version 3 stuff has been "re-engineered" to conform to the OSI 7-layer spesifications and protocol requirements. There is no "OSI Network Program" per-se. All the "new" programs are "OSI Network Programs" while all the "old" stuff are the "Proprietary Network Programs" > 6. If the answers to these questions are not resolved in my favor, does > anyone want to buy a loaded 3B1? (Or, maybe a 6386. It isn't clear > which "incompatible" hardware to dump.) (Or^^2, where is the OSI > support for the 1,000's of 3B1s??) The 386 software and hardware is far superrior for use on the STARLAN and STARLAN 10 networks. Their capacity (in network connections/sessions) is more than twice that for the 3B2/600 et. al. If and when the SCSI adapter comes out for the 386 systems, they will become the network server of choice. At present 3B2/600 is the best mass-disk server while a 386 server will handle the MS-DOS clients better. A multiple service network using one STARLAN-DOS server and an RFS link seem to be the best ideas for combining the capacities of the two. As far as the 3B1 are concerned, the best answer I have gotten on that is "that hardware isnolonger supported." Oh well..... Rob. Disclaimer: This is my research, not "official" AT&T party line; The first, however, is damn close to the second. > -- > Paul Homchick {allegra | rutgers | uunet} !cbmvax!cgh!paul > Chimitt Gilman Homchick, Inc.; One Radnor Station, Suite 300; Radnor, PA 19087