From cls at rkey.com Tue Feb 11 10:13:16 2003 From: cls at rkey.com (Craig Simon) Date: Tue, 11 Feb 2003 13:13:16 -0500 Subject: [ih] European and Asian participation around 1989 Message-ID: <3E493D3C.1070202@rkey.com> Hi all, I recently heard about an incident in which a number of European and Asian networks briefly lost their connections to the Internet in 1989 or 1990. Someone in the US Government apparently decided that the registry functions then being sponsored through the DOD should not be used to support services for connections outside the US. This presumably accelerated the process of transferring responsibility for the funding of registry services to the NSF. Is this story on the mark at all? Can anyone here clarify the background of this? Thanks, Craig Simon From aca at cs.utexas.edu Tue Feb 25 07:37:55 2003 From: aca at cs.utexas.edu (Adriana C. Arrington) Date: Tue, 25 Feb 2003 09:37:55 -0600 (CST) Subject: [ih] Dec 1969 meeting and Telnet In-Reply-To: <200212200445.gBK4j3C08286@boreas.isi.edu> References: <200212200445.gBK4j3C08286@boreas.isi.edu> Message-ID: I need clarification for the meeting in December of 1969 when Larry Roberts "redirected" the implementers to try a more layered approach (RFC 1000, p 4). What happened at that meeting? Were there any minutes kept from that meeting? >From what I can tell in RFC 15 and RFC 1000, it seems that the first rejected version of the network protocols of December 1969 was not layered at all. For instance, the "Telnet" at that time encapsulated everything between the transport and application layers, as we know them today. So then this solution was not broad enough requiring the "redirection", layering and the inventions of Host-Host and the next version of Telnet. So did the network look something like this (based on RFC 15): ----------------------------- | telnet | random | text | | | compiler| editor | ----------------------------- | OS with interface | | to Host-IMP layer | ----------------------------- instead of the layered design of Telnet, ICP, and Host-Host, which came a year or so later?? What kind of asymmetry was in this first set of protocols that is different than the asymmetry of Old Telnet (see RFC 1000 p.4)? Thanks again, Adriana mailto:aca at cs.utexas.edu http://www.cs.utexas.edu/~aca From steve at stevecrocker.com Tue Feb 25 07:51:16 2003 From: steve at stevecrocker.com (Steve Crocker) Date: Tue, 25 Feb 2003 10:51:16 -0500 Subject: [ih] RE: Dec 1969 meeting and Telnet In-Reply-To: Message-ID: <002101c2dce5$bb789aa0$ada8a8c0@SCROCKER> In December 1969 we were in an awkward state. The first few IMPs had been installed and we didn't have a protocol suite ready. We had rigged a simple telnet-like protocol between UCLA and SRI as a demo. Also, the SRI guys had rigged a way to pass files back and forth to Utah in an ad hoc fashion, although I'm unsure of the timing; it may have been later. We kept groping for the right primitives to use as the base layer and we hadn't quite settled on it. Feeling pressured by the existence of the IMPs and no host level software, we proposed to Larry Roberts that we defer the general approach and simply build a telnet protocol directly. (I think that's accurate; Vint or others may have a different recollection.) Larry firmly responded that he wanted to see the generality and could accept the delay. There weren't any formal minutes, but I don't recall whether someone jotted this down in an RFC. I don't quite understand your diagram, but perhaps the above answers your question. Steve > -----Original Message----- > From: Adriana C. Arrington [mailto:aca at cs.utexas.edu] > Sent: Tuesday, February 25, 2003 10:38 AM > To: internet-history at postel.org > Cc: steve at stevecrocker.com; vinton.g.cerf at wcom.com; Chris > Edmondson-Yurkanan; Adriana C. Arrington > Subject: Dec 1969 meeting and Telnet > > > I need clarification for the meeting in December of 1969 when > Larry Roberts "redirected" the implementers to try a more > layered approach (RFC 1000, p 4). What happened at that > meeting? Were there any minutes kept from that meeting? > > >From what I can tell in RFC 15 and RFC 1000, it seems that > the first rejected version of the network protocols of > December 1969 was not layered at all. For instance, the > "Telnet" at that time encapsulated everything between the > transport and application layers, as we know them today. So > then this solution was not broad enough requiring the > "redirection", layering and the inventions of Host-Host and > the next version of Telnet. > > So did the network look something like this (based on RFC 15): > > ----------------------------- > | telnet | random | text | > | | compiler| editor | > ----------------------------- > | OS with interface | > | to Host-IMP layer | > ----------------------------- > > instead of the layered design of Telnet, ICP, and Host-Host, > which came a year or so later?? > > What kind of asymmetry was in this first set of protocols > that is different than the asymmetry of Old Telnet > (see RFC 1000 p.4)? > > Thanks again, > Adriana > mailto:aca at cs.utexas.edu http://www.cs.utexas.edu/~aca From day at std.com Tue Feb 25 10:29:17 2003 From: day at std.com (John Day) Date: Tue, 25 Feb 2003 13:29:17 -0500 Subject: [ih] RE: Dec 1969 meeting and Telnet In-Reply-To: <00a601c2dcf9$5b844e00$ada8a8c0@SCROCKER> References: <00a601c2dcf9$5b844e00$ada8a8c0@SCROCKER> Message-ID: At 13:11 -0500 2/25/03, Steve Crocker wrote: >Yes, that's exactly right. It was quite impressive. I didn't mean to >slight their work by saying they passed files back and forth. Didn't think you were. I wasn't even sure if I remembered that right! ;-) > >Now that I think about it, I recall Jeff Rulifson saying the process was >*faster* than doing it entirely locally on the SDS 940 machine because >it didn't burden their poor disk as much. That is, it was faster to >send the file to Utah and get back the result than it was to read and >write on the local disk. I forget the rest of the details, but it's >probably worth dredging up. ;-) that is great! Do remember why they were doing it? Was it that Utah had a Tenex and SRI was getting theirs and in the meantime were compiling stuff in Utah? Can't dredge that up. A year or so later, we were developing an OS for a PDP-11 and were compiling in San Diego and downloading to the 11 in Illinois. Then all we could do was flip switches at the console or force a dump to the printer! From day at std.com Tue Feb 25 09:52:16 2003 From: day at std.com (John Day) Date: Tue, 25 Feb 2003 12:52:16 -0500 Subject: [ih] RE: Dec 1969 meeting and Telnet In-Reply-To: <002101c2dce5$bb789aa0$ada8a8c0@SCROCKER> References: <002101c2dce5$bb789aa0$ada8a8c0@SCROCKER> Message-ID: At 10:51 -0500 2/25/03, Steve Crocker wrote: >In December 1969 we were in an awkward state. The first few IMPs had >been installed and we didn't have a protocol suite ready. We had rigged >a simple telnet-like protocol between UCLA and SRI as a demo. Also, the >SRI guys had rigged a way to pass files back and forth to Utah in an ad I have a vague recollection that SRI-NLS were cross-compiling on the machine at Utah and then using the code files back at SRI. Is that right? Take care, John >hoc fashion, although I'm unsure of the timing; it may have been later. >We kept groping for the right primitives to use as the base layer and we >hadn't quite settled on it. Feeling pressured by the existence of the >IMPs and no host level software, we proposed to Larry Roberts that we >defer the general approach and simply build a telnet protocol directly. >(I think that's accurate; Vint or others may have a different >recollection.) Larry firmly responded that he wanted to see the >generality and could accept the delay. > >There weren't any formal minutes, but I don't recall whether someone >jotted this down in an RFC. > >I don't quite understand your diagram, but perhaps the above answers >your question. > >Steve > > > > > > -----Original Message----- > > From: Adriana C. Arrington [mailto:aca at cs.utexas.edu] > > Sent: Tuesday, February 25, 2003 10:38 AM > > To: internet-history at postel.org > > Cc: steve at stevecrocker.com; vinton.g.cerf at wcom.com; Chris > > Edmondson-Yurkanan; Adriana C. Arrington > > Subject: Dec 1969 meeting and Telnet > > > > > > I need clarification for the meeting in December of 1969 when > > Larry Roberts "redirected" the implementers to try a more > > layered approach (RFC 1000, p 4). What happened at that > > meeting? Were there any minutes kept from that meeting? > > > > >From what I can tell in RFC 15 and RFC 1000, it seems that > > the first rejected version of the network protocols of > > December 1969 was not layered at all. For instance, the > > "Telnet" at that time encapsulated everything between the > > transport and application layers, as we know them today. So > > then this solution was not broad enough requiring the > > "redirection", layering and the inventions of Host-Host and > > the next version of Telnet. > > > > So did the network look something like this (based on RFC 15): > > > > ----------------------------- > > | telnet | random | text | > > | | compiler| editor | > > ----------------------------- > > | OS with interface | > > | to Host-IMP layer | > > ----------------------------- > > > > instead of the layered design of Telnet, ICP, and Host-Host, > > which came a year or so later?? > > > > What kind of asymmetry was in this first set of protocols > > that is different than the asymmetry of Old Telnet > > (see RFC 1000 p.4)? > > > > Thanks again, > > Adriana > > >mailto:aca at cs.utexas.edu >http://www.cs.utexas.edu/~aca From steve at stevecrocker.com Tue Feb 25 10:11:45 2003 From: steve at stevecrocker.com (Steve Crocker) Date: Tue, 25 Feb 2003 13:11:45 -0500 Subject: [ih] RE: Dec 1969 meeting and Telnet In-Reply-To: Message-ID: <00a601c2dcf9$5b844e00$ada8a8c0@SCROCKER> Yes, that's exactly right. It was quite impressive. I didn't mean to slight their work by saying they passed files back and forth. Now that I think about it, I recall Jeff Rulifson saying the process was *faster* than doing it entirely locally on the SDS 940 machine because it didn't burden their poor disk as much. That is, it was faster to send the file to Utah and get back the result than it was to read and write on the local disk. I forget the rest of the details, but it's probably worth dredging up. Steve > -----Original Message----- > From: John Day [mailto:day at std.com] > Sent: Tuesday, February 25, 2003 12:52 PM > To: Steve Crocker; 'Adriana C. Arrington'; internet-history at postel.org > Cc: vinton.g.cerf at wcom.com; 'Chris Edmondson-Yurkanan'; > 'Steve Crocker' > Subject: [ih] RE: Dec 1969 meeting and Telnet > > > At 10:51 -0500 2/25/03, Steve Crocker wrote: > >In December 1969 we were in an awkward state. The first few > IMPs had > >been installed and we didn't have a protocol suite ready. We had > >rigged a simple telnet-like protocol between UCLA and SRI as > a demo. > >Also, the SRI guys had rigged a way to pass files back and forth to > >Utah in an ad > > I have a vague recollection that SRI-NLS were cross-compiling on the > machine at Utah and then using the code files back at SRI. Is that > right? > > Take care, > John > > >hoc fashion, although I'm unsure of the timing; it may have > been later. > >We kept groping for the right primitives to use as the base > layer and > >we hadn't quite settled on it. Feeling pressured by the > existence of > >the IMPs and no host level software, we proposed to Larry > Roberts that > >we defer the general approach and simply build a telnet protocol > >directly. (I think that's accurate; Vint or others may have > a different > >recollection.) Larry firmly responded that he wanted to see the > >generality and could accept the delay. > > > >There weren't any formal minutes, but I don't recall whether someone > >jotted this down in an RFC. > > > >I don't quite understand your diagram, but perhaps the above answers > >your question. > > > >Steve > > > > > > > > > > > -----Original Message----- > > > From: Adriana C. Arrington [mailto:aca at cs.utexas.edu] > > > Sent: Tuesday, February 25, 2003 10:38 AM > > > To: internet-history at postel.org > > > Cc: steve at stevecrocker.com; vinton.g.cerf at wcom.com; Chris > > > Edmondson-Yurkanan; Adriana C. Arrington > > > Subject: Dec 1969 meeting and Telnet > > > > > > > > > I need clarification for the meeting in December of 1969 when > > > Larry Roberts "redirected" the implementers to try a more > > > layered approach (RFC 1000, p 4). What happened at that > > > meeting? Were there any minutes kept from that meeting? > > > > > > >From what I can tell in RFC 15 and RFC 1000, it seems that > > > the first rejected version of the network protocols of > > > December 1969 was not layered at all. For instance, the > > > "Telnet" at that time encapsulated everything between the > > > transport and application layers, as we know them today. So > > > then this solution was not broad enough requiring the > > > "redirection", layering and the inventions of Host-Host and > > > the next version of Telnet. > > > > > > So did the network look something like this (based on RFC 15): > > > > > > ----------------------------- > > > | telnet | random | text | > > > | | compiler| editor | > > > ----------------------------- > > > | OS with interface | > > > | to Host-IMP layer | > > > ----------------------------- > > > > > > instead of the layered design of Telnet, ICP, and Host-Host, > > > which came a year or so later?? > > > > > > What kind of asymmetry was in this first set of protocols > > > that is different than the asymmetry of Old Telnet > > > (see RFC 1000 p.4)? > > > > > > Thanks again, > > > Adriana > > > > >mailto:aca at cs.utexas.edu > >http://www.cs.utexas.edu/~aca > From steve at stevecrocker.com Tue Feb 25 10:31:58 2003 From: steve at stevecrocker.com (Steve Crocker) Date: Tue, 25 Feb 2003 13:31:58 -0500 Subject: [ih] RE: Dec 1969 meeting and Telnet In-Reply-To: Message-ID: <00a701c2dcfc$2ef0fd90$ada8a8c0@SCROCKER> I don't recall the motivation. Jeff Rulifson would likely know. I think he's still at Sun. Steve > -----Original Message----- > From: John Day [mailto:day at std.com] > Sent: Tuesday, February 25, 2003 1:29 PM > To: Steve Crocker; 'John Day'; 'Adriana C. Arrington'; > internet-history at postel.org > Cc: vinton.g.cerf at wcom.com; 'Chris Edmondson-Yurkanan' > Subject: RE: [ih] RE: Dec 1969 meeting and Telnet > > > At 13:11 -0500 2/25/03, Steve Crocker wrote: > >Yes, that's exactly right. It was quite impressive. I > didn't mean to > >slight their work by saying they passed files back and forth. > > Didn't think you were. I wasn't even sure if I remembered > that right! ;-) > > > > >Now that I think about it, I recall Jeff Rulifson saying the process > >was > >*faster* than doing it entirely locally on the SDS 940 > machine because > >it didn't burden their poor disk as much. That is, it was faster to > >send the file to Utah and get back the result than it was to read and > >write on the local disk. I forget the rest of the details, but it's > >probably worth dredging up. > > ;-) that is great! Do remember why they were doing it? Was it that > Utah had a Tenex and SRI was getting theirs and in the meantime were > compiling stuff in Utah? Can't dredge that up. > > A year or so later, we were developing an OS for a PDP-11 and were > compiling in San Diego and downloading to the 11 in Illinois. Then > all we could do was flip switches at the console or force a dump to > the printer! > From touch at ISI.EDU Tue Feb 25 11:24:21 2003 From: touch at ISI.EDU (Joe Touch) Date: Tue, 25 Feb 2003 11:24:21 -0800 Subject: [ih] test - please ignore Message-ID: <3E5BC2E5.9090705@isi.edu> This is a test to internet-history at postel.org Please ignore. Joe From doleary at juniper.net Tue Feb 25 15:18:27 2003 From: doleary at juniper.net (dave o'leary) Date: Tue, 25 Feb 2003 15:18:27 -0800 Subject: Fwd: Re: [ih] European and Asian participation around 1989 Message-ID: <5.1.1.6.2.20030225151702.020bd5c8@zircon.juniper.net> Per Craig's suggestion. I would not be surprised if others have alternate recollections; I was not sleeping a whole lot in those days... :-) dave >Date: Tue, 25 Feb 2003 10:58:26 -0500 >From: Craig Simon >To: "dave o'leary" >Subject: Re: [ih] European and Asian participation around 1989 > >Hi, > >Thanks very much for your response. > >I've been picking up bits and pieces here and there about the early >internationalization of the Internet, but no confirmation yet of the >"thrown off" incident. > >It seems that your answer might be of interest to people on the IH list. >Since you already made the effort of writing it, I wonder if you would >consider reposting it there as a response to my thread. This might have >the side benefit of prompting further responses from others. > >In any case, thanks again. > >Craig > >dave o'leary wrote: >>At 10:37 AM 2/21/2003 -0500, Craig Simon wrote: >> >>>Hi, >>> >>>Thanks for responding, >>> >>>There was no answer from anyone on the list. I've contacted at least >>>three other people off the list who I thought might be able to confirm >>>what Don Mitchell had told me, but they had no recollection of this. >>> >>>Anything you can add would be huge help. >> >>In the fall of 1989 I went to work as the technical manager >>at SURAnet, based in College Park, Maryland. We hosted >>an NSFnet node on campus at the University of Maryland. >>In the spring of 1990 we added ESnet (DoE backbone) >>and NSI (NASA backbone) nodes, and established >>FIX-East. We also had an ARPAnet connection, but it >>was decommissioned along with the remaining nodes >>and connections. >>In the summer of 1990, we moved to a building just off >>campus, and in addition to the other connections listed >>above, we added a Milnet connection and other links >>including what was called the "UK Fat Pipe". This was >>a multiplexed T1 which ran from University College London >>to NASA Goddard where some of the channels (768kb?) >>were dropped off and then the rest of the channels were >>terminated on a router connected to FIX-East. I know that >>this wasn't the first Internet connection from the US to the UK, >>but I know that once it came up (pretty sure this was mid-1990) >>we didn't bring it back down for any policy reasons as >>described below. The reason that the connection was split >>between Goddard and FixEast is because there were multiple >>agencies involved in paying for the connection, but I don't >>remember which ones. I left SURAnet in late March of 1992, >>and the UK connection was definitely still in place. >>I definitely remember the registry funding conversations, >>but I don't remember any time that connectivity was dropped. >>The scenario you describe might have happened earlier in 1989, >>and may have been the motivation for the multi-agency UK Fat >>Pipe. Tony Hain who is now at Cisco or Steve Wolff (who I think >>is also still at Cisco) might know more. Tony was responsible >>for engineering at ESnet back then and Steve was the program >>manager at NSF. Also Milo Medin who was running the NSI >>backbone and had responsibility for FIX West. >>In addition to the UK connection at FIX-East, there was also >>another connection from Germany, but I can't remember where >>it terminated in the US. >>Connectivity to Asia was pretty sparse in those days, >>only Japan and Australia I think. Jun Marai (Japan) and >>Geoff Huston who built AARnet and the initial US connectivity >>would remember the specific dates of when their connections >>came up. I do remember that FIX-West at NASA Ames came >>up a little before FIX-East, because we were moving our >>facility off campus, so the Japan and Australia connections >>might have come up early enough to in turn be taken down >>for some policy reason. >>I'm pretty sure that I have some network maps at my house >>somewhere, and some notes from meetings, etc. back in those >>days, which might help sort through some of this. >>I hope this helps; as I think about this I am beginning to >>recollect more... let me know if I can help you track down >>any of the folks above. >> dave >> >>>Craig >>> >>>dave o'leary wrote: >>> >>>>At 01:13 PM 2/11/2003 -0500, you wrote: >>>> >>>>>Hi all, >>>>> >>>>>I recently heard about an incident in which a number of European and >>>>>Asian networks briefly lost their connections to the Internet in 1989 >>>>>or 1990. Someone in the US Government apparently decided that the >>>>>registry functions then being sponsored through the DOD should not be >>>>>used to support services for connections outside the US. This >>>>>presumably accelerated the process of transferring responsibility for >>>>>the funding of registry services to the NSF. >>>>> >>>>>Is this story on the mark at all? Can anyone here clarify the >>>>>background of this? >>>> >>>> >>>>Did you receive any responses on this? If so, I am interested >>>>in a summary. >>>>I can provide insight from my perspective as a network operator >>>>back then, but I doubt I will capture any behind the scenes political >>>>intrigue. >>>> dave From gih at telstra.net Tue Feb 25 20:52:31 2003 From: gih at telstra.net (Geoff Huston) Date: Wed, 26 Feb 2003 15:52:31 +1100 Subject: Fwd: Re: [ih] European and Asian participation around 1989 In-Reply-To: <5.1.1.6.2.20030225151702.020bd5c8@zircon.juniper.net> Message-ID: <5.1.0.14.2.20030226154526.01f6d008@localhost> from the dusty archive of my mind dept: >>>Connectivity to Asia was pretty sparse in those days, >>>only Japan and Australia I think. and New Zealand >>> Jun Marai (Japan) and >>>Geoff Huston who built AARnet and the initial US connectivity >>>would remember the specific dates of when their connections >>>came up. New Zealand to Hawaii - around March 1989 14.4 analogue modem using a cable circuit to Hawaii Australia - around August 1989, using a 56Kbps satellite circuit to Hawaii Japan around the same time, but I can't recall the month. The Australian circuit was never 'taken down" although at one stage we were informed that the DoD routers were running close to table capacity and we (AARNet) had to list those address prefixes where there was active research activities that included a requirement to interface in to the DoD networks. This was done and DoD connectivity for those that knew that they needed it was never affected. Our connections to other federal research networks and our transit services to other non-US Internet networks were never affected. regards, Geoff Huston From braden at ISI.EDU Wed Feb 26 12:12:59 2003 From: braden at ISI.EDU (Bob Braden) Date: Wed, 26 Feb 2003 20:12:59 GMT Subject: [ih] Re: internet-history digest, Vol 1 #120 - 2 msgs Message-ID: <200302262012.UAA09963@gra.isi.edu> *> >>In the summer of 1990, we moved to a building just off *> >>campus, and in addition to the other connections listed *> >>above, we added a Milnet connection and other links *> >>including what was called the "UK Fat Pipe". This was *> >>a multiplexed T1 which ran from University College London *> >>to NASA Goddard where some of the channels (768kb?) *> >>were dropped off and then the rest of the channels were *> >>terminated on a router connected to FIX-East. I know that *> >>this wasn't the first Internet connection from the US to the UK, See RFC 831 for information on the earliest Internet connection to the UK. Bob Braden From braden at ISI.EDU Wed Feb 26 12:25:11 2003 From: braden at ISI.EDU (Bob Braden) Date: Wed, 26 Feb 2003 20:25:11 GMT Subject: [ih] Re: internet-history digest, Vol 1 #120 - 2 msgs Message-ID: <200302262025.UAA09980@gra.isi.edu> *> >>>>> *> >>>>>I recently heard about an incident in which a number of European and *> >>>>>Asian networks briefly lost their connections to the Internet in 1989 *> >>>>>or 1990. Someone in the US Government apparently decided that the *> >>>>>registry functions then being sponsored through the DOD should not be *> >>>>>used to support services for connections outside the US. This *> >>>>>presumably accelerated the process of transferring responsibility for *> >>>>>the funding of registry services to the NSF. This probably refers to the end of SATNET, DARPA's experimental satellite network, which happened around then, I believe. (A quick check of the IAB minutes around that time did not find a reference to the demise of SATNET.) SATNET connected the US to our principle European collaborators in TCP/IP development, the UK, Norway, and West Germany. Bob Braden From cls at rkey.com Thu Feb 27 07:52:08 2003 From: cls at rkey.com (Craig Simon) Date: Thu, 27 Feb 2003 10:52:08 -0500 Subject: [ih] NACRS? Message-ID: <3E5E3428.7040804@rkey.com> Hello again, NACRS were evidently used in the 1980s and early 1990s as a kind of template submitted to SRI by email by individuals requesting connection to the network. Is this correct? Does anyone recall what the acronym stands for or anything about the procedures by which NACRS were typically used? Thanks, Craig Simon From srh at merit.edu Thu Feb 27 15:45:45 2003 From: srh at merit.edu (Susan Harris) Date: Thu, 27 Feb 2003 18:45:45 -0500 (EST) Subject: [ih] NACRS? In-Reply-To: <3E5E3428.7040804@rkey.com> Message-ID: > NACRS were evidently used in the 1980s and early 1990s as a kind of > template submitted to SRI by email by individuals requesting connection > to the network. Is this correct? Does anyone recall what the acronym > stands for or anything about the procedures by which NACRS were > typically used? During the NSFNET the regional networks used Network Announcement/Change Requests to update the AS690 route table. NACRs were gracefully replaced by RADB route objects when the NSFNET went away in 1995. From doleary at juniper.net Thu Feb 27 18:52:01 2003 From: doleary at juniper.net (dave o'leary) Date: Thu, 27 Feb 2003 18:52:01 -0800 Subject: [ih] NACRS? In-Reply-To: References: <3E5E3428.7040804@rkey.com> Message-ID: <5.1.1.6.2.20030227184331.059b4d28@zircon.juniper.net> At 06:45 PM 2/27/2003 -0500, Susan Harris wrote: > > NACRS were evidently used in the 1980s and early 1990s as a kind of > > template submitted to SRI by email by individuals requesting connection > > to the network. Is this correct? Does anyone recall what the acronym > > stands for or anything about the procedures by which NACRS were > > typically used? > >During the NSFNET the regional networks used Network Announcement/Change >Requests to update the AS690 route table. NACRs were gracefully >replaced by RADB route objects when the NSFNET went away in 1995. Just to be a bit more clear - initially EGP and post cutover, the BGP protocol was used to update the routing tables. NACRS were used to update to policy database, reflecting which advertisements should be accepted by which regional and international networks. Since there were plenty of sites that were dual homed, and EGP did not provide much semantic for indicating which path should be preferred for a given destination, the policy database reflected which advertisement should be considered primary/secondary/tertiary for a given prefix when it was advertised into the backbone in multiple locations. The NACR consisted of an electronic form which had to be submitted by one of the designated individuals at a given regional network. I don't remember the details but I believe someone at Merit contacted the various regional folks directly when a network under their purview changed status (so the policy for a given prefix couldn't be changed without buy-in from the others involved). I don't remember the exact dates when these began, but it wasn't until the early 90's. In the mid and late 80's with the original FuzzBall based NSFnet, we approximately coordinated all this manually. Mark Oros (then at Cornell) was one of the early coordinators of the policies, which were at that time mostly implemented via gated.conf files. dave