[e2e] on local ethernet throughput?
Cannara
cannara at attglobal.net
Tue Oct 23 10:47:43 PDT 2001
David, perhaps I misread your original words, but as you say, few coax
segments exist today. However, every hub has a segment as its backplane, and
10/100/1000 ones have as many as 3, interconnected by bridging/switching
processors. So, my main concern is that people understand where collisions
occur, where they don't, and how to assess their importance. That last has
been one of the religious issues you mention that has unfortunately been
misunderstood and exploited, even today, not only by competitive media folks,
but by equipment vendors scaring folks to buy gear they don't really need.
One thing we must all keep in mind is that the more complex the gear, the more
chance of failure. It's no accident that a major car assembler in the midwest
ran for a dozen years on original DEC-installed Ethernet coax, 24hrs/day; and
was being extremely careful, a few years back, when assessing the value and
reliability of newer hubs and switch products -- no network, no robots, no
cars, no $.
Alex
"David P. Reed" wrote:
>
> At 09:55 PM 10/21/2001 -0700, Cannara wrote:
> >This is a good msg in that it exemplifies misconceptions that we all should
> >have mastered:
> >
> >1) A "hub" is not an arbitrator, it is simply a MAC extender (originally, more
> >correctly called a "multiport repeater").
>
> When I said "not your grandfather's Ethernet" I was including trivial hubs
> in the "grandfather's Ethernet" class, Alex. The hubs I was referring to
> are the modern 10/100 hubs, which as you point out, are essentially
> semi-switches.
>
> And my main point stands: all of the gear you can buy today does not fit
> the original coax-based form of the collision domain model, which made
> worst case assumptions about how collisions happen that are not possible in
> today's "Ethernet" (TM) networks.
>
> I think you are agreeing with this point. Aside from the above confusion,
> I'm not sure where you get the idea I am propagating misconceptions, though.
>
> - David
> --------------------------------------------
More information about the end2end-interest
mailing list