[ih] 13 the unlucky number

Vint Cerf vint at google.com
Thu Aug 13 01:13:02 PDT 2020


Jack, wasn't that
[2] Strazisar, V., "Gateway Routing: An Implementation Specification,"
IEN-30, Bolt Beranek and Newman Inc., August 1979.
[3] Strazisar, V., "How to Build a Gateway," IEN-109, Bolt Beranek and
Newman Inc., August 1979.

v

On Wed, Aug 12, 2020 at 9:23 PM Jack Haverty via Internet-history <
internet-history at elists.isoc.org> wrote:

> Hi Barbara,
>
> AFAIK, there was never a spec for gateway/router behavior, except Ginny
> Strazisar's very early RFC.  However, the behavior of ICMP Echo
> Request/reply et al was spec'd in RFC 792 in 1981, so the gateway should
> have been conforming to that.   If the destination host was down, it
> should have returned a "Destination Unreachable".
>
> Part of the problem may have resulted from a little glitch in the DoD's
> process of declaring the Internet protocols to be DoD Standards.  When
> they did that in the early 80s, they spec'd both the TCP and IP RFCs as
> standards, but overlooked the ICMP spec.   I remember we started seeing
> machines come online with "DoD Standard implementations" as specified by
> their companies contracts with whatever piece of DoD ordered the
> equipment.   They had implemented TCP and IP, but not ICMP.  After much
> complaining by us at BBN and probably others, ICMP eventually got added
> in some way, but it may have been a while before that spec actually got
> into products.
>
> I have only anecdotal evidence, but it seems things have regressed over
> the decades and now routers don't always respond to ICMP.  I think some
> ISPs have turned it off because of all the "ping" traffic they
> experienced.   I guess they don't consider it part of the Standard either.
>
> I remember when we decided at some meeting to split out the ICMP
> functionality from the basic IP definition, mostly just for
> organizational simplification.   It's amazing to me what kinds of
> longstanding and far reaching consequences a little mistake (overlooking
> ICMP as a critical component of TCP/IP) can have.
>
> /Jack
>
>
> On 8/12/20 5:58 PM, Barbara Denny via Internet-history wrote:
> >  I have a ping story.  I got sent to Germany in the mid-80s  (1986?) to
> help install an IP testbed for USAREUR.  We used the Cisco AGS routers
> (The ones with the fan that sounded like an airplane taking off ).  Cisco
> was a very small company so we got a lot of support in the bug fixing
> department.  I remember having to call them one night from Germany because
> I didn't like what their ping implementation was doing.  I noticed their
> routers would respond to a ping even if  the interface with the ping's IP
> destination address was down.  I didn't like that behavior so I think I
> ended up having to get Kirk Lougheed? on the line to get them to agree to
> change this for us. I've always wondered if the behavior of ping on a
> gateway /router was specified.  I certainly didn't have the time, or
> resources at my finger tips,  to check this out at the time.
> >
> > barbara
> >
> >     On Wednesday, August 12, 2020, 4:54:15 PM PDT, Jacob Goense via
> Internet-history <internet-history at elists.isoc.org> wrote:
> >
> >  The fuzzball archive seems down at the moment, but it is still
> > there[1]. Mills' PING was a client and server, or "pinger and
> > sounder" as he called it. Muus' ping(8) was the client side to
> > complement the BSD kernel.
> >
> >
> http://web.archive.org/web/20131020063249/http://malarky.udel.edu/~dmills/data/du0/PING.MAC
> >
> > On 2020-08-12 07:01, Louis Mamakos via Internet-history wrote:
> >> The timing here lines up with Dave Mills' fuzzball code we had
> >> running at University of Maryland (where Dave was a visiting
> >> professor for a "special topics" networking class.)  I do
> >> fondly recall Fuzzball PING and it was certainly a well-used
> >> tool for the initial debugging of the UNIVAC TCP/IP stack that
> >> Mike Petry and I started implementing at that time - fall 1980.
> >>
> >> It was also around that time that we added the initial Ethernet
> >> support in the Fuzzball code, adding ARP and a QBus InterLAN
> >> interface.  Once again, PING was the universal debugging tool.
> >>
> >> I do recall seeing "Packet InterNet Groper" in the Fuzzball
> >> source code.
> >>
> >> louie
> >>
> >> On 11 Aug 2020, at 16:21, Jack Haverty via Internet-history wrote:
> >>
> >>> Ping may have been released in 1983 but it was in use long before
> >>> that.    When I was in charge of keeping the "core gateways" running,
> >>> Dave Mills was famous for doing lots of experiments that often gave us
> >>> heartburn.   I clearly recall him telling us at some Internet meeting
> >>> about his experiments and the tool he used - he called it "Ping", and
> >>> explained it was an acronym for "Packet InterNet Groper".  This was
> >>> probably 1979/80 or thereabouts.  I don't know that Dave invented
> >>> "ping", but I believe that's where I first heard about it.
> >>>
> >>> /Jack
> >>>
> >>>
> >>> On 8/11/20 1:06 PM, Alejandro Acosta via Internet-history wrote:
> >>>> uh, good question.
> >>>>
> >>>> Based on the days of the firsts networks and the release of ping the
> >>>> answer is none, afaik, ping was released in 1983
> >>>>
> >>>> On 8/11/20 2:43 PM, Jack Haverty via Internet-history wrote:
> >>>>> A related question, if you're exploring Internet History, might be
> >>>>> "Which of the early networks were ever actually operational nets on
> >>>>> The
> >>>>> Internet, i.e., nets that you could ping and get a response?"
> >>>>>
> >>>>> I was involved in the 77-80s timeframe, and as I recall, many of
> >>>>> those
> >>>>> low numbered networks were assigned numbers, but didn't actually
> >>>>> ever
> >>>>> get connected to the operational Internet.
> >>>>>
> >>>>> /Jack Haverty
> >>>>>
> >>>>> On 8/11/20 10:53 AM, Alejandro Acosta via Internet-history wrote:
> >>>>>> Hello All,
> >>>>>>
> >>>>>>    First, really thanks for your comments.
> >>>>>>
> >>>>>>    I read a little bit more about the network 13. I supposed I
> >>>>>> should
> >>>>>> have done this before sending the email.
> >>>>>>
> >>>>>>    As I said, it does not appear in RFC 790 (Sep 81), and it does
> >>>>>> not
> >>>>>> appear until RFC 990 (Nov 1986 assigned to XEROX)
> >>>>>>
> >>>>>>    However, I just realized that actually network 13 was first seen
> >>>>>> in
> >>>>>> RFC 739 assigned to National Physical Laboratory and last seen in
> >>>>>> RFC
> >>>>>> 776.
> >>>>>>
> >>>>>>
> >>>>>> Thanks again & sorry for the noise.
> >>>>>>
> >>>>>>
> >>>>>> Alejandro,
> >>>>>>
> >>>>>>
> >>>>>> On 8/11/20 12:45 PM, Alex McKenzie via Internet-history wrote:
> >>>>>>>    Alejandro,
> >>>>>>> I don't think any of us can speak for Jon Postel, who assigned the
> >>>>>>> numbers, and sadly he is no longer with us to speak for himself.
> >>>>>>> I
> >>>>>>> knew Jon pretty well and he showed no evidence of being a
> >>>>>>> superstitious person.  I think Steve Crocker's explanation that
> >>>>>>> the
> >>>>>>> number was assigned to an entity that could not yet be made public
> >>>>>>> on
> >>>>>>> the date RFC 790 was released is the most likely answer.
> >>>>>>> For what its worth,Alex McKenzie
> >>>>>>>
> >>>>>>>       On Tuesday, August 11, 2020, 9:08:58 AM EDT, Alejandro
> >>>>>>> Acosta
> >>>>>>> via Internet-history <internet-history at elists.isoc.org> wrote:
> >>>>>>>      Hello list,
> >>>>>>>
> >>>>>>>      I have a question and one more time I believe this a good
> >>>>>>> place
> >>>>>>> to ask.
> >>>>>>>
> >>>>>>>      During the weekend I read the old RFC 790 (ASSIGNED NUMBERS).
> >>>>>>> When
> >>>>>>> reading it I noticed the following:
> >>>>>>>
> >>>>>>> {...}
> >>>>>>>
> >>>>>>>          009.rrr.rrr.rrr   BRAGG-PR      Ft. Bragg Packet Radio
> >>>>>>> Net
> >>>>>>> [JEM]
> >>>>>>>          010.rrr.rrr.rrr   ARPANET       ARPANET [17,1,VGC]
> >>>>>>>          011.rrr.rrr.rrr   UCLNET        University College London
> >>>>>>> [PK]
> >>>>>>>          012.rrr.rrr.rrr   CYCLADES      CYCLADES [VGC]
> >>>>>>>          013.rrr.rrr.rrr                 Unassigned [JBP]
> >>>>>>>          014.rrr.rrr.rrr   TELENET       TELENET [VGC]
> >>>>>>>          015.rrr.rrr.rrr   EPSS          British Post Office EPSS
> >>>>>>> [PK]
> >>>>>>>          016.rrr.rrr.rrr   DATAPAC       DATAPAC [VGC]
> >>>>>>>          017.rrr.rrr.rrr   TRANSPAC      TRANSPAC [VGC]
> >>>>>>>          018.rrr.rrr.rrr   LCSNET        MIT LCS Network
> >>>>>>> [43,10,DDC2]
> >>>>>>>
> >>>>>>> {...}
> >>>>>>>
> >>>>>>>
> >>>>>>>      As you can see the 013.rrr.rrr.rrr was unassigned but some
> >>>>>>> subsequent
> >>>>>>> prefix were (014, 015 ..... ). Is there any reason for it?. I know
> >>>>>>> 013
> >>>>>>> was later assigned to XEROX-NET.
> >>>>>>>
> >>>>>>>      I wonder if 013 was skipped because some sort of
> >>>>>>> superstitions?.
> >>>>>>>
> >>>>>>>
> >>>>>>> Thanks,
> >>>>>>>
> >>>>>>>
> >>>>>>> Alejandro,
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>> -- Internet-history mailing list
> >>> Internet-history at elists.isoc.org
> >>> https://elists.isoc.org/mailman/listinfo/internet-history
>
> --
> Internet-history mailing list
> Internet-history at elists.isoc.org
> https://elists.isoc.org/mailman/listinfo/internet-history
>


-- 
new postal address:
Google, LLC
1900 Reston Metro Plaza, Suite 1400
Reston, VA 20190


More information about the Internet-history mailing list