joe@eng3.UUCP (Joe LaRocque) (07/24/90)
I have been given a 'chance to excell' by my boss. Simply put, how do we go about getting a base EtherNet Address assigned to us? I seem to recall that PARC is still in charge of these numbers. But, I know that they have moved to San Diego and I no longer have a name or telephone number for an individual that I can talk to about this request. Before I forget....I know that we could get a set of proms from a mfg who would take care of the problem for us. Our problem is that the new system we are building requires as few surface mount structures as possible, so we will be assigning the EtherNet Address via software. Thanks for your assist! Joe
stewart@xyplex.com (Bob Stewart) (07/25/90)
Joe (and anyone else who's interested), The most recent communication I had from Xerox got me some protocol types. As of March 1990, the address was: Xerox Corporation Xerox Systems Institute 475 Oakmead Parkway Sunnyvale, CA 94086 (408) 737-4652 The letter was signed by Fonda Lix Pallone, Customer Administration. That should get you to the right neighborhood. Bob ----------- Bob Stewart (rlstewart@eng.xyplex.com) Xyplex, Boxborough, Massachusetts (508) 264-9900
oberman@rogue.llnl.gov (07/25/90)
In article <488@eng3.UUCP>, joe@eng3.UUCP (Joe LaRocque) writes: > I have been given a 'chance to excell' by my boss. Simply put, how do > we go about getting a base EtherNet Address assigned to us? I seem to > recall that PARC is still in charge of these numbers. But, I know that > they have moved to San Diego and I no longer have a name or telephone > number for an individual that I can talk to about this request. > > Before I forget....I know that we could get a set of proms from a mfg > who would take care of the problem for us. Our problem is that the new > system we are building requires as few surface mount structures as > possible, so we will be assigning the EtherNet Address via software. > I believe that the IEEE now hand out these numbers, although Xerox still does the actual work. But I am concerned with the idea that you are planning on putting out a device which gets it's Ethernet address from software. I don't have the Ethernet or 802.3 spec handy, but I believe that this is NOT legal. And, even if it is, it's dangerous. It is critical that all Ethernet devices have globally unique addresses. The hardware assignment of these ainsures that there can NEVER be two the same. The portion of the spec allowing software to reset this address is something I've always objected to, but it is there. R. Kevin Oberman Lawrence Livermore National Laboratory Internet: oberman@icdc.llnl.gov (415) 422-6955 Disclaimer: Don't take this too seriously. I just like to improve my typing and probably don't really know anything useful about anything.
kevinr@moe.Tandem.COM (Kevin J. Rowett) (07/26/90)
In article <1990Jul25.090203.1@rogue.llnl.gov>, oberman@rogue.llnl.gov writes: |> In article <488@eng3.UUCP>, joe@eng3.UUCP (Joe LaRocque) writes: |> |> I don't have the Ethernet or 802.3 spec handy, but I believe that this is NOT |> legal. And, even if it is, it's dangerous. It is critical that all Ethernet |> devices have globally unique addresses. The hardware assignment of these |> ainsures that there can NEVER be two the same. The portion of the spec allowing |> software to reset this address is something I've always objected to, but it is |> there. The IEEE spec makes a distintion between globally admin'ed 802.3 addresses and locally admin'ed address. It's actually a bit in the address field. The caveat is that you can have some assurance global admin'ed addresses won't conflict between vendors, but all bets are off if the address is locally admin'ed. In products we build, we give the board a default globally admin'ed address. We still allow the end user to change the current address, but the software will never let him pick a globally assigned style address. Unfortunetly, that aprt of the S/W hasn't been popular with our customers. kevinr@Tandem.com
postel@VENERA.ISI.EDU (07/26/90)
Hi. Ethernet Address blocks are now assigned by the IEEE (212-705-7092). Ethernet Types are still assigned by Xerox (408-737-4652). --jon.
montague@npdiss1.StPaul.NCR.COM (John Montague) (07/26/90)
In article <488@eng3.UUCP> joe@eng3.UUCPJoe LaRocque writes: >I have been given a 'chance to excell' by my boss. Simply put, how do >we go about getting a base EtherNet Address assigned to us? I seem to By this request I believe you mean "How does one obtain an assignment of a block of Universal LAN addresses?" These addresses are assigned by the IEEE Standards Office, 445 Hoes Lane, Piscataway, NJ 08854-4150; there is a nominal fee ($1000). >Before I forget....I know that we could get a set of proms from a mfg >who would take care of the problem for us. Our problem is that the new >system we are building requires as few surface mount structures as >possible, so we will be assigning the EtherNet Address via software. Universally administered addresses must be uniquely assigned to a single LAN node. This normally accomplished by making a permanent address assignment to a physical assembly in a non-volitile register (NOT battery backed-up RAM). If you wish to assign addresses through software you should use "locally administered addresses" which you may choose to be similar to the address you have permanaently assigned to the hardware, differing only in the "Address Administration" bit (the second most significant bit in the 48 bit address: 0= universally administered, 1= locally administered). Extreem care must be taken to ensure that duplicate addresses NEVER occur on the network. John Montague W0RUE Manager, Standards & Architecture NCR, Network Products Division, St. Paul, MN john.montague@stpaul.NCR.COM
kevinr@moe.Tandem.COM (Kevin J. Rowett) (07/26/90)
In article <146@npdiss1.StPaul.NCR.COM>, montague@npdiss1.StPaul.NCR.COM (John Montague) writes: |> |> Universally administered addresses must be uniquely assigned to a single Gosh the "I" in IEEE sure has grown. I = institue ( as in it ain't a std till I say so) I = International ( as in globally admin addresses) I = Interplanetary ( as in universal) Anyone got the MAC address of the HST? kevinr@tandem.com N6RCE
rpw3@rigden.wpd.sgi.com (Rob Warnock) (07/27/90)
In article <1990Jul25.184017.192@tandem.com> kevinr@Tandem.COM writes: +--------------- | In products we build, we give the board a default globally admin'ed | address. We still allow the end user to change the current address, | but the software will never let him pick a globally assigned | style address. +--------------- Notice that if are are running the Xerox NS (XNS) protocols on a multi-homed host or gateway, you need to be able to set the physical address of all Ethernet interfaces in the host to the same value, as XNS uses that address as the host address. That is, in XNS the globally-unique "host address" is a property of the *host*, not of the interface(s) as in TCP/IP. [Reference: "Internet Transport Protocols", Xerox System Integration Standard XSIS-028112, page 16] What is often to simplify assignment of the "host address" is to use the physical (globally administered) address off the "primary" network interface, or if there is no reasonable notion of "primary", all the physical addresses are read at boot time and the smallest/largest/one_in_slot#0/etc. is chosen as "the" host address, and then that value is written back into *all* of the network interfaces. Clearly, your software restriction would prevent this. +--------------- | Unfortunately, that part of the S/W hasn't been popular with our customers. +--------------- I wonder why... ;-} -Rob ----- Rob Warnock, MS-9U/510 rpw3@sgi.com rpw3@pei.com Silicon Graphics, Inc. (415)335-1673 Protocol Engines, Inc. 2011 N. Shoreline Blvd. Mountain View, CA 94039-7311
lindh@uhasun.hartford.edu (Andrew Lindh) (07/28/90)
In article <1990Jul25.090203.1@rogue.llnl.gov>, oberman@rogue.llnl.gov writes > ....... > putting out a device which gets it's Ethernet address from software. > > I don't have the Ethernet or 802.3 spec handy, but I believe that this is NOT > legal. And, even if it is, it's dangerous. It is critical that all Ethernet > devices have globally unique addresses. The hardware assignment of these > ainsures that there can NEVER be two the same.The portion of the spec allowing > software to reset this address is something I've always objected to, but it is > there. Every DEC computer uses a software address!!! If you want to run DECnet you have to change your address with software. I do not know if DEC computers also have a hardware address. In this environment it is very simple to have two computers with the same address...it drives our bridges crazy! -- Andrew Lindh, a student at the University of Hartford -- Computer Science INTERNET: lindh@uhasun.hartford.edu | NOTE: All views here are MINE!!! BITNET: lindh@hartford.bitnet | Not the school's or those of anyone else! UUCP: lindh@uhasun.uucp | ---- When will I graduate??? "SYNFU!"
kevinr@moe.Tandem.COM (Kevin J. Rowett) (07/28/90)
In article <65311@sgi.sgi.com>, rpw3@rigden.wpd.sgi.com (Rob Warnock) writes: |> In article <1990Jul25.184017.192@tandem.com> kevinr@Tandem.COM writes: |> | address. We still allow the end user to change the current address, |> | but the software will never let him pick a globally assigned |> | style address. |> +--------------- |> as "the" host address, and then that value is written back into *all* of the |> network interfaces. |> |> Clearly, your software restriction would prevent this. Not really, as the customer can set ther MAC address to any locally admin'ed address they choose. kevinr@Tandem.com
rpw3@rigden.wpd.sgi.com (Rob Warnock) (07/28/90)
In article <1990Jul27.232750.22472@tandem.com> kevinr@Tandem.COM writes: +--------------- | In article <65311@sgi.sgi.com>, rpw3@rigden.wpd.sgi.com (Rob Warnock) writes: | |> In article <1990Jul25.184017.192@tandem.com> kevinr@Tandem.COM writes: | |> | address. We still allow the end user to change the current address, | |> | but the software will never let him pick a globally assigned | |> | style address. | |> +--------------- | |> as "the" host address, and that value is written back into *all* of the | |> network interfaces. | |> Clearly, your software restriction would prevent this. | Not really, as the customer can set ther MAC address to any locally admin'ed | address they choose. +--------------- But that completely negates all of the advantages of the known-unique globally-administered addresses which come "for free" on each network interface! If you allow setting network physical addresses to any value [which, by the way, nearly all of the hardware interfaces do], no customer ever has to assign/hand_out/dis-ambiguate XNS addresses. All of the addresses on all of the network boards on all of the hosts in his internet are globally unique, and the "host addresses" that get used are merely a subset of the globally-unique numbers available on each host. [Single-interface hosts will default to the GU MAC address of their single interface, as usual.] In what you propose, all of a sudden the *customer* is responsible for not assigning any duplicate addresses ANYWHERE IN HIS/HER INTERNET, which may span half the world and, which is worse, many different administrative boundaries within the curtomer's own organization. To me, this is a *major* loss of convenience and reliability, simply because someone in your company decided that "customers [really, customer's network software] can't be trusted to do it right." PLEASE! We don't need any more "well-meaning" paternalism! -Rob ----- Rob Warnock, MS-9U/510 rpw3@sgi.com rpw3@pei.com Silicon Graphics, Inc. (415)335-1673 Protocol Engines, Inc. 2011 N. Shoreline Blvd. Mountain View, CA 94039-7311