[ih] Internet-history Digest, Vol 57, Issue 4

[Craig Partridge]

Acknowledging that John S had a better viewpoint as things happened (I
showed up in the later part of the Ethernet vs. token ring vs. token bus
[Ungermann Bass?] debates), I think Ethernet did four things right and then
benefitted from one wonderful invention.

The four right things were:

   - big enough addresses (not obvious at the time, witness IPv4's too
   small).
   - media access rules that were contention based and worked for both
   wired and wireless (again, not obvious, but paved the way for Ethernet
   framing to work on WiFi)
   - big enough error check -- CRC32 is amazing and has lasted even as
   error patterns have evolved dramatically (may just be showing its age, but
   only just)
   - media access rules that allowed the channel to be filled. &ou can
   transmit at 9.X megabits on a 10 megabit Ethernet - a point Dave Boggs made
   with great vigor in his SIGCOMM '88 talk, in response to the folks who had
   claimed Ethernet could only operate at 1/e capacity (so no one pushed for a
   "more efficient" protocol)

The wonderful invention was the learning bridge, which made it possible to
stick Ethernet's together without hand configuration.  Probably our first
great case of autoconfiguring network deployments and all kudos to Radia
for making it happen.

Craig

On Thu, Aug 22, 2024 at 1:11 PM John Shoch via Internet-history <
internet-history at elists.isoc.org> wrote:

> John Levine wrote:
>
> "It's quite impressive that forty years later Ethernet has the same
> logical formats as the Ethernet that ran on thick yellow coax, even
> though at the hardware level practically nothing is the same."
>
> In retrospect, the most important and enduring aspect of the DIX Ethernet
> spec (now evolved through multiple generations of design over multiple
> decades) was the packet format.
>
> At Xerox we learned a lot from the1st generation implementation and
> operation of the Experimental Ethernet (2.94 Mbps) and the Pup internet
> architecture -- a combination which had well-matched but small address
> fields.
>
> James Carville advised Bill Clinton that, "It's the economy, stupid!"
> I often observe, "it's the addressing, dummy!"
>
> While the first generation worked well, with thousands of machines and
> dozens of networks, it became clear it would not continue to scale.  It
> required a much broader view of addressing, which led to the audacious idea
> of 48-bit absolute addresses --- that's what allowed the 2nd generation DIX
> Etherent and XNS internet to scale together.
>
> The person we should thank for that is Yogen Dalal:
>
> http://www.bitsavers.org/pdf/xerox/parc/techReports/OPD-T8101_48-Bit_Absolute_Internet_and_Ethernet_Host_Numbers.pdf
> https://dl.acm.org/doi/10.1145/800081.802680
>
> John Shoch
>
> PS:  If you read the paper, you will see that the 48-bit absolute addresses
> were intended for and implemented on both  the DIX Ethernet and as part of
> XNS internet architecture -- a well crafted "impedance match."  (Other
> people will have a much more informed view on the later, continuing
> migration from IPv4 to IPv6.....)
>
> >
> > Message: 4
> > Date: 21 Aug 2024 18:06:47 -0400
> > From: "John Levine" <johnl at iecc.com>
> > To: internet-history at elists.isoc.org
> > Subject: Re: [ih] "This is the History of Ethernet."
> > Message-ID: <20240821220647.6AD3392357F0 at ary.lan>
> > Content-Type: text/plain; charset=utf-8
> >
> > It appears that Brian E Carpenter via Internet-history <
> > brian.e.carpenter at gmail.com> said:
> > >BTW, it's worth noting that the *current* IETF STD 37 for ARP (a.k.a.
> RFC
> > 826) says:
> > >
> > >"This protocol was originally designed for the DEC/Intel/Xerox
> > >10Mbit Ethernet.  It has been generalized to allow it to be used
> > >for other types of networks.  Much of the discussion will be
> > >directed toward the 10Mbit Ethernet."
> >
> > It's quite impressive that forty years later Ethernet has the same
> > logical formats as the Ethernet that ran on thick yellow coax, even
> > though at the hardware level practically nothing is the same.
> >
> > The Ethernets in my house are all gigabit twisted pair, and if I upgrade
> > it'll likely be to 10G fiber.  But ARP still works.
> >
> > R's,
> > John
> >
> >
> --
> Internet-history mailing list
> Internet-history at elists.isoc.org
> https://elists.isoc.org/mailman/listinfo/internet-history
>


-- 
*****
Craig Partridge's email account for professional society activities and
mailing lists.