Why we need a length field in Ethernet while we don't need it in FDDI??

More homework?

Clearly, school has started for the Winter Quarter/Spring Semester. (Seeing that the poster is from Hong Kong, it may be his Summer Quarter/Fall Semester.)

That said, we *don't* need a length field in Ethernet. The fact that the vast majority of Ethernet frames ever sent (e.g., those carrying TCP/IP payloads) do not contain an Ethernet length field speaks volumes.

What I find more amazing is that, in this day and age, networking instructors are still teaching FDDI and Token Ring technology as if was something *necessary* for the emerging professional to understand (other than for historical, "how did we get here" reasons.) I agree that: (1) it is useful to understand the various ways in which a shared-access medium can be arbitrated, and (2) new media may indeed necessitate the use of such access methods (or new ones based on the same principles).

However, what I generally see is instruction that is more in tune with the 1980s, when shared-media LANs were dominant, and there was market competition among systems using different access methods. This is no longer the case. Part of the problem may be that most existing texts dedicate a lot of pages to this subject; the teacher is either loath to dispense with it, or doesn't have anything else prepared to take its place.

When I teach LAN technology, I discuss shared-media access and the tradeoffs among access methods in a more "abstract" manner, i.e., what system characteristics would lead you to use contention access vs. polling/token-passing; the "details" of 802.3 vs. 802.4 vs. 802.5 vs. FDDI are used merely as *examples* of those tradeoffs, not as "important distinctions" among competing technologies.

Also (for what it's worth), you can watch a streaming-video sample of a LAN class I taught a few years back, at:

-- Rich Seifert Networks and Communications Consulting 21885 Bear Creek Way (408) 395-5700 Los Gatos, CA 95033 (408) 228-0803 FAX

Send replies to: usenet at richseifert dot com

A friend says he still has one arcnet network at a client! As for token ring, FDDI & classfull IP, they should only be taught for historical background. Any instructor who presents them as current technolgy needs to have his course updated. Even IBM's given up on token ring.

This sort of reminds me of my high school electronics class (many years ago). Virtually the entire course was about vacuum tubes, with only a brief look at transistors. A year later, I was working with integrated circuits, which hadn't even been mentioned in that electronics class. The kids who are preparing for tomorrow seem to get stuck being taught about yesterday's technology.

:> What I find more amazing is that, in this day and age, networking :> instructors are still teaching FDDI and Token Ring technology as if was :> something *necessary* for the emerging professional to understand

:Well they teach classfull IP Networking (and only classfull) too.

It also turns out that in some circumstances, the Cisco PIX software will default netmasks according to the IP class.. even in cases where the IP is in the subnet assigned to an interface and so the netmask would be easily deducable.

There is also a circumstance under which the Cisco VPN software will get told a classfull subnet mask by any of Cisco's devices that support IPSec, such as the PIX, VPN Concentrator, or IOS with IPSec Feature Set.

Well they teach classfull IP Networking (and only classfull) too. And even today you can find TokenRing Networks, we just pulled out the last components in December.

Jens

You don't

I can just see him getting that question marked wrong, because the instructor instists a length field is necessary. Perhaps Rich should have a word with the instructor.

As an experiment, that student should run a network monitor on a network, to see what's actually on the wire. He'll likely see ethernet II, which doesn't have a length field. In fact, I recommend he examine all the info in some frames, so he can see how the lessons fit in the real world.

You've just pointed out two of the major problems with education as it is practiced in the US today. First, you get good grades by giving the answer the instructor wants, not the answer that is correct, and second, there is at best minimal effort expended showing how it fits into the real world.

We are still running some systems that identify themselvs as 'SYSTEM V/68 Release R3V6 Version 990318 M68030' and the Copyright is Copyright (c) 1984 AT&T Copyright (c) 1985 Motorola, Inc. They still work in terms of the old Classes!