ARPANET Technical Information: Packet Formats, Maps, etc


Introduction

The ARPANET is to this century, and the next several, what Gutenberg's press was to the last several: an invention which generated a major quantum change in society - one of only a few such in human history (others being the invention of writing, and of the telegraph/telephone).

Information technology is a 'force multiplier' for other all other technologies, because it increases their effects many times over. In the technology world, information technologies are to other technologies what Gauss famously opined of mathematics - that it was the Queen of the sciences.

Technical information about the ARPANET can be difficult to locate online, unless you know to search for magic strings like '1822' and 'NIC 8246'. There is a very good, and quite accurate, book about the ARPANET, Where Wizards Stay up Late, by Katie Hafner, but it lacks technical detail. There are three very good technical papers which give a great deal of detail:

  F.E. Heart, R.E. Kahn, S.M. Ornstein,  W.R. Crowther, and D.C. Walden,
        "The interface message processor for the ARPA computer network",
        Proceedings AFIPS, 1970 SJCC, Vol. 36, pp. 551-567.

  J.M. McQuillan, W.R. Crowther, B.P. Cosell, D.C. Walden, and F.E. Heart,
        "Improvements in the Design and Performance of the ARPA Network",
        Proceedings AFIPS, 1972 FJCC, Vol. 40, pp. 741-754.

  John M. McQuillan, David C. Walden, "The ARPA Network Design Decisions",
	in "Computer Networks", Vol. 1, No. 5, August 1977, pp. 243-289.
but those are not available freely online.

This site is an attempt to rectify this situation to some degree. It contains some useful technical information (packet formats, maps, etc) along with links to online technical resources. (It has been split up into multiple pages since there are so many images.)

Note: The rest of this page assumes you know something about the ARPANET. So if you don't know what an IMP is, you need to do some other reading first to get some background.


Maps

BB&N produced maps of the ARPANET on a regular basis, as it was always changing (usually growing, until its last years when it was slowly dismantled and turned off). They were produced in both 'geographic' and 'logical' form.

Here are a substantial selection of both types (on separate pages, to reduce the overall page size):

There is also this interesting animated image, which shows a sequence of early geographic maps.

MILNET maps

In 1982, the DoD decided to split the ARPANET up into two networks: one, called MILNET, would be an operational network for use by the military; the other, to retain the name ARPANET, would be used by the research community. The community was formally notified of this plan in December, 1982.

The plan was to group all the hosts on the ARPANET into two sets, and then use the access control features in the IMP software to isolate each group from the other; hosts in one group would only be able to send packets to other hosts in their group. Access from one group to the other would happen via 'mail bridges', routers which were 'attached' to both sets.

Once the network had been 'virtually' split into two networks in this manner, work could begin on actually separating it physically - assigning each IMP to one network or the other, and deleting all links which led from an 'ARPANET' IMP to a 'MILNET' IMP.

The first phase of this split happened in October, 1983; the second phase was not completed until September, 1984. Prior to the split, in 1983, there were 113 IMPs in the ARPANET; after the ARPANET/MILNET split, the MILNET consisted of 65 nodes, leaving the ARPANET with 68 nodes.

For historical interest, a few sets of MILNET maps are included here.


Packet formats

There were two different kinds of packet formats used in the ARPANET: those used for "packets" sent between IMPs (these are more like 'frames' in modern parlance, since a single user "message" might be broken up into several "packets" for transmission through the network), and those used for packets (technically, "messages") sent between IMPs and their connected hosts.

There were two versions of the Host-IMP headers:

You can find the packet formats laid out in detail here.


Online technical material

These are original ARPANET technical documents, provided by major institutions, and thus unlikely to suffer link-rot. (The amount of things which one can find links to, but which '404' when you click on the links, is amazing. Hopefully these will not suffer from that problem.)
Alas, the final major piece of NCP (NIC #7101, "Offical Initial Connection Protocol") does not appear to be online at this point (although see this early version, which is almost identical to the final version).

In addition to these, the early RFC's (available ubiquitously across the Internet) are also full of interesting technical material; RFC's 1 through about 700 are almost entirely ARPANET related - only after about 700 does Internet material start to appear. Most are quite short; the entire collection (less the few long ones) can be quickly scanned in an evening. (This page may eventually contain pointers to the most important ones.)


Back to JNC's home page

© Copyright 2012 by J. Noel Chiappa

Last updated: 15/May/2012