Archived from groups: comp.dcom.lans.ethernet (
More info?)
fran_beta@hotmail.com wrote:
>
> J. Clarke wrote:
>> fran_beta@hotmail.com wrote:
>>
>> > I'm trying to mark a paper that deals with networks and I'm trying to
>> > establish whether a question is correct.
>>
>> When you say you're "trying to mark a paper" do you mean "grade a test"?
>
>
> Yes
>
>> If
>> so, then the "right answer" would be the one that matches what the
>> students were told in class and in their texts--anything else would be
>> unfair and if what they were told in class and in the texts is wrong then
>> you need to address that rather than punishing the students for the
>> faculty's error.
>
> Correct again, but this paper contains the ambiguity described and if
> there is no correct and sensible answer, or if there is more than one
> sensible answer I am ethically bound to deem all such answers correct.
>
> Originally, I examined the paer and found the term "hub" where I said
> there was a gateway. I assumed this was an error and replaced it so
> that the ethernet linked-devices could exchange data with the token
> ring devices, but with hindsight, the "error" might have been
> intentional -- an attempt to psoe this very question and test knowledge
> of the difference between a gateway and a hub.
>
> I've found this a little confusing myself as gateways and "bridges" are
> often used interchangeably in terminological terms at least. A
> "gateway" seems to be a more sophisticated bridge, and a "router" may
> well contain hardware components that allow it to move data between
> different protocols.
Generally speaking, a "bridge" connects two networks at the datalink layer
(the layer that defines "ethernet" or "token ring" or "arcnet" or
whatever), a router operates at the network or transport layer (the layer
that defines "TCP/IP", "IPX/SPX", etc), and a gateway operates above this
level.
Any literature that refers to a bridge as a gateway or vice versa is _not_
to be trusted--it is either very sloppy or very old.
>> If you are a grader or a TA then you really should ask the professor this
>> question.
>>
>
> I'm a teacher working in a setting where this coterie here is likely to
> be the best qualified to answer.
>
>> If you mean that you are trying to answer a homework question then you
>> really should learn to research this sort of thing yourself, however
>> since the question seems to be ill-posed in this case I can forgive a
>> certain confusion.
>>
>
>
> Indeed
>
>> > The question offers a diagram of a network topology bringing together
>> > ethernet (star and bus) and token rings (includes a printer). A number
>> > of devices are attached to each of the networks. These networks are
>> > joined by a gateway.
>>
>> Right here there's a problem. "Gateway" is a very fuzzy term. It could
>> be
>> a bridge, a router, or something at a higher level than that. The
>> students are going to have to make some assumptions about the definition
>> of "gateway"--the sharp ones will see this and state their assumptions,
>> the less sharp ones will go on whatever notion they may have of its
>> definition assuming that what they know is correct.
>>
>
>
>
>
>> > The question is whether any of the following statements is correct.
>> >
>> > Not all data packets pass through the gateway (My guess, since the
>> > token ring only passes packets to those with the correct token)
>>
>> Leaving aside the nature of ring networks, since the gateway is between
>> the Token Ring and something else, the nature of Token Ring has little to
>> do
>> with what gets passed through the gateway. That will be determined by
>> the specific nature of the gateway--all, some, or none may be passed.
>>
>
> So does that mean that the question as posed is impossible to answer
> with certainty?
You are correct.
> I assumed that it was a feature of token rings that any data packet
> passing within the ring (eg between a node and the attached printer)
> would not pass through the gateway.
If the gateway is attached to the token ring then each packet will be seen
by the gateway--what the gateway does with it is another story.
>> > Data packets are sent only to the intended receiver (the ethernet
>> > broadcasts to all nodes, so this can't be true)
>>
>> It can be if all the Ethernet and Token Ring ports are on a common bridge
>> (such bridges do exist).
>>
>
> |||||||||||||
>
> The diagram showed:
>
>
> Area 1
>
> Standard Star Network
>
> Devices:
>
> Hub; MinPC; Disk Array; IBM AS/400
AS/400? Now _that_ complicates things, and the effect depends on the
vintage--early AS/400s could not communicate directly with PC
printers--something emulating the kind of printer or print server it
expects had to be attached to the network--I don't know to what extent this
has changed in more recent models.
> a LAN cable of some sort joins it to the gateway/(hub in original)
That might be Twinax--the AS/400 has its own unique network standard, called
"System Application Architecture" (different from SNA, lurkers) used
primarily to connect terminals and printers. That would make sense as SAA
can have hubs, although there may be IBM-ese for them that I don't recall.
In any case with the AS/400 a gateway makes more sense--it would likely be
an SAA gateway--a couple of examples are Novell "Netware for SAA" and
Microsoft "SNA Server", which also supports SAA. Among other things it
would probably expose PC printers to the AS/400 and provide terminal
emulation, and may provide other services.
At one time, before IBM repositioned their midrange and mainframe systems as
servers, this was one of the the standard ways to connect an AS/400 to a PC
network, the other being to put 5250 (the standard AS/400 terminal was the
IBM 5250) emulation boards in the PCs that were to be used as terminals.
I'm no AS/400 expert by the way, you might want to try to find a more
specific newsgroup for advice on that end--it's really outside of the
Ethernet world and the PC world as well--it's an IBM Midrange system, which
used to be considered "heavy iron", and it's better to think of it as a
mainframe than as a PC.
> Two LAN cables leave the gateway one to the:
>
>
> Token Ring in Area 2
>
> Devices: Laser Printer; IBM PC; Server;
>
>
> and the other to:
>
> Area 3 Standard Bus Network
>
> Devices: iMAC; server; Printer; raid drive
"Standard bus network" doesn't mean anything. In fact there isn't any bus
network to which an iMac can be attached directly--its ports are 100TX
IIRC--you'd need a converter to attach to 10base2 and I don't know if it
works with localtalk at all.
> ||||||||||||||
>
>
>> > All data packets are transmitted to all devices (again the token ring
>> > networks fails this)
>>
>> Well, actually, you seem to have a misconception about Token Ring. In a
>> Token Ring each device sends data to one and only one device, the one on
>> the other end of the cable connected to the transmitter in its NIC. The
>> MAU is an essentially passive device.
>>
>
> OK
>
>> If the ring is not switched, then the data goes from one machine to the
>> next around the ring until it gets to its destination, and the
>> destination then retransmits the data around the ring until it returns to
>> the originator,
>> which compares what was sent with what was received. So each frame
>> actually passes through _all_ devices on a non-switched Token Ring. If
>> it is switched then the switch sends the frame to its destination
>> directly.
>>
>
> I see, so technically, the data packet does go to each device but is
> rejected until it reaches the one for which it was intended?
Not rejected, just passed along and otherwise ignored.
>> > Printing cannot be initiated from the devices in the star network (this
>> > must be wrong since a gateway joins the ring and star)
>>
>> Depends on the design of the gateway. If it is one that either supports
>> printing directly or passes the necessary packets then printing can take
>> place, otherwise it can't.
>>
>
> You seem to be saying that there isn't sufficient information to say.
Exactly.
>> > Would the answer be different if there were a hub rather than a gateway
>> > (I assume this would divide the token ring from the rest of the
>> > network)?
>>
>> A "hub" is, compared to a "gateway", a well defined device--these days
>> with many devices sold as "hubs" actually being bridges it isn't as well
>> defined
>> as it once was, but it's still much clearer than "gateway". Since there
>> is no such thing as a "hub" that will join Ethernet and Token Ring, they
>> would indeed be isolated.
>>
>>
>
> That's what I thought.
>
>
> Thanks
>
> Fran
--
--John
to email, dial "usenet" and validate
(was jclarke at eye bee em dot net)