Token ring and ethernet on same physical network? - TCP-IP

This is a discussion on Token ring and ethernet on same physical network? - TCP-IP ; hello "subnetting allows a single physical network to be logically divided regardless of the physical layout of a network, since it is possible to divide a physical network into several subnets by configuring different host computers to use different routers." ...

+ Reply to Thread
Results 1 to 12 of 12

Thread: Token ring and ethernet on same physical network?

  1. Token ring and ethernet on same physical network?

    hello

    "subnetting allows a single physical network to be logically divided
    regardless of the physical layout of a network, since it is possible
    to divide a physical network into several subnets by configuring
    different host computers to use different routers."

    I can see how that is possible if network ( which we logically divided
    into two subnetworks S1 and S2) has star topology, since HUB can send
    the incoming packet intended for S1 through the ports which connect to
    the computers residing on S1.

    But I don't understand how we can, with the help of subnetting,
    logically divide Token Ring and Ethernet network.
    For starters, how can Ethernet and Token ring exists together as one
    physical network?
    Perhaps there is a device similar to HUB which uses some ports to
    connect Token ring computers and other ports connect Ethernet
    computers --> the device then based on subnet ID decides through which
    ports will the incoming packet go?! But other than that, I don't
    understand how else it would eb possible!


    thank you

    cheers


  2. Re: Token ring and ethernet on same physical network?

    On Jun 27, 12:39 pm, kaja_love...@yahoo.com wrote:

    > "subnetting allows a single physical network to be logically divided
    > regardless of the physical layout of a network, since it is possible
    > to divide a physical network into several subnets by configuring
    > different host computers to use different routers."


    This is a misleading statement, though it is literally true. Perhaps
    it makes more sense in context.

    > I can see how that is possible if network ( which we logically divided
    > into two subnetworks S1 and S2) has star topology, since HUB can send
    > the incoming packet intended for S1 through the ports which connect to
    > the computers residing on S1.


    It has nothing to do with where the hubs send the packets. The packets
    go to their destinations somehow, that's all that matters.

    > But I don't understand how we can, with the help of subnetting,
    > logically divide Token Ring and Ethernet network.
    > For starters, how can Ethernet and Token ring exists together as one
    > physical network?


    They can't. The statement above stars with the assumption that we're
    dealing with a single physical network and says nothing about
    subnetting in multiple physical networks (which is probably more
    common).

    > Perhaps there is a device similar to HUB which uses some ports to
    > connect Token ring computers and other ports connect Ethernet
    > computers --> the device then based on subnet ID decides through which
    > ports will the incoming packet go?! But other than that, I don't
    > understand how else it would eb possible!


    You are confusing two things that have nothing to do with each other.
    Using multiple subnets on the same physical network has nothing to do
    with using multiple subnets on different physical networks. The
    statement above was talking about the first case and you are talking
    about the second.

    DS


  3. Re: Token ring and ethernet on same physical network?

    wrote:

    > "subnetting allows a single physical network to be logically divided
    > regardless of the physical layout of a network, since it is possible
    > to divide a physical network into several subnets by configuring
    > different host computers to use different routers."
    >
    > I can see how that is possible if network ( which we logically divided
    > into two subnetworks S1 and S2) has star topology, since HUB can send
    > the incoming packet intended for S1 through the ports which connect to
    > the computers residing on S1.
    >
    > But I don't understand how we can, with the help of subnetting,
    > logically divide Token Ring and Ethernet network.
    > For starters, how can Ethernet and Token ring exists together as one
    > physical network?


    In principle, they can. In principle, you can bridge between Ethernet
    and Token Ring, or Ethernet and FDDI, or Ethernet, Token Ring, and ATM
    with LANE, or between different flavors of Ethernet (such as optical
    stars and coax buses). Whether such bridges are successful in the
    marketplace is a completely different question, of course. We're just
    talking about how the standards are written.

    So you can create a single IP subnet which consists of different types
    of layer 2 networks. In principle. If you do not stick an IP router
    between these different L2 nets, all the hosts conncected to these nets
    could belong to the same IP subnet.

    > Perhaps there is a device similar to HUB which uses some ports to
    > connect Token ring computers and other ports connect Ethernet
    > computers --> the device then based on subnet ID decides through which
    > ports will the incoming packet go?! But other than that, I don't
    > understand how else it would eb possible!


    The device you're speaking of exists: it's called a "bridge."

    How does the bridge decide where to "route" the frames? Same as any L2
    "switch" does. Initially, it floods frames to all ports. Then, it learns
    which MAC addresses live at each port, and it more cleverly "routes"
    frames only where they are wanted.

    And no, it does not need to look at the IP layer at all. Not unless you
    want to do clever tricks, like IGMP snooping and such.

    Bert


  4. Re: Token ring and ethernet on same physical network?

    "Vernon Schryver" wrote:

    >>In principle, they can. In principle, you can bridge between Ethernet
    >>and Token Ring, or Ethernet and FDDI, or Ethernet, Token Ring, and ATM
    >>with LANE, or between different flavors of Ethernet (such as optical
    >>stars and coax buses). Whether such bridges are successful in the
    >>marketplace is a completely different question, of course. We're just
    >>talking about how the standards are written.

    >
    > That is wrong, except for fraudulent marketoon definitions of "can
    > bridge." The fatal problem is that the 802.3 MTU is only 1500 bytes
    > while both 802.5 Token Ring and FDDI want larger frames.


    That's hardly "fatal," if you consider upper layer techniques like Path
    MTU Discovery. And also, the more typical case is, and was, that
    Ethernet was used to connect to the clients, so that whatever occurred
    over FDDI or ATM/LANE would typically have been limited to 1500 bytes
    anyway.

    Still, it's just wrong to say that these L2 nets can't be bridged, to
    someone who is trying to understand the underlying concepts. We're not
    trying to push useless merchandise here. I'm totally ignoring marketing,
    and these products are not on store shelves anymore anyway.

    You make perfectly valid, practical points. However, the OP's questions
    were conceptual in nature. He's trying to understand what makes these
    network protocols tick.

    Bert


  5. Re: Token ring and ethernet on same physical network?

    On Jun 28, 3:04 pm, "Albert Manfredi"
    wrote:
    > "Vernon Schryver" wrote:
    > >>In principle, they can. In principle, you can bridge between Ethernet
    > >>and Token Ring, or Ethernet and FDDI, or Ethernet, Token Ring, and ATM
    > >>with LANE, or between different flavors of Ethernet (such as optical
    > >>stars and coax buses). Whether such bridges are successful in the
    > >>marketplace is a completely different question, of course. We're just
    > >>talking about how the standards are written.

    >
    > > That is wrong, except for fraudulent marketoon definitions of "can
    > > bridge." The fatal problem is that the 802.3 MTU is only 1500 bytes
    > > while both 802.5 Token Ring and FDDI want larger frames.

    >
    > That's hardly "fatal," if you consider upper layer techniques like Path
    > MTU Discovery. And also, the more typical case is, and was, that
    > Ethernet was used to connect to the clients, so that whatever occurred
    > over FDDI or ATM/LANE would typically have been limited to 1500 bytes
    > anyway.



    Unfortunately PMTU discovery doesn't work in this scenario. There's
    no one to send back the fragmentation required ICMP message, unless
    you have the bridge do it, and that's going to be a nasty hack.

    Source routed token-ring could sort-of deal with that, if the network
    topology was not too complex, but required the end node on the TRN
    side to support SR, which was rare, at best, for IP.


  6. Re: Token ring and ethernet on same physical network?

    wrote:

    >> That's hardly "fatal," if you consider upper layer techniques like
    >> Path
    >> MTU Discovery. And also, the more typical case is, and was, that
    >> Ethernet was used to connect to the clients, so that whatever
    >> occurred
    >> over FDDI or ATM/LANE would typically have been limited to 1500 bytes
    >> anyway.

    >
    >
    > Unfortunately PMTU discovery doesn't work in this scenario. There's
    > no one to send back the fragmentation required ICMP message, unless
    > you have the bridge do it, and that's going to be a nasty hack.


    True, if there's no router in the path at all. But a fairly normal
    situation would be that the FDDI or ATM or mixed net would be a
    backbone, and the clients would be on access subnets. In which case,
    you'd have a router that could send these ICMP replies back to some
    heavy duty backbone machine trying to huge large frames.

    I guess Token Rings were used for client connections, so PMTU itself
    would not work so well. But, the TCP MSS option should work anyway, even
    without a router in the path.

    Bert


  7. Re: Token ring and ethernet on same physical network?

    On Jun 28, 9:24 pm, v...@calcite.rhyolite.com (Vernon Schryver) wrote:
    > You can bridge 802.5 rings into one
    > network. You can also bridge FDDI rings into one network. >

    (...) You also run into problems bridging 802.5
    > and FDDI.



    Well, memories are growing dim, but IIRC bridging 4Mb TRN and FDDI was
    OK, since they had the same maximum frame size, but 16Mb TRN bumped
    the frame size to 17800. Just to add to the confusion... ;-)


  8. Re: Token ring and ethernet on same physical network?

    In article ,
    Albert Manfredi wrote:

    >> Albert Manfredi's claim about bridging is like saying that you can
    >> use diesel, gasoline, ethenol, and methanol in cars and rejecting
    >> the objections of people who don't have multi-fuel engines or have
    >> fuel systems that methanol dissolves as minor or irrelevant.

    >
    >A passable analogy.
    >
    >If someone who is learning about internal combustion piston engines asks
    >whether these can operate with fuels other than gasoline, a person who
    >misses the forest through the trees will raise all sorts of objections.
    >Without appreciating that the objections are secondary to the initial
    >question.


    No one who cares about the pure theory of internal combustion piston
    engines would ask whether a car can use methanol. Those who care about
    the pure, academic theory need to be told about _unspecified_ fuel,
    oxidizer, and (most inert) gas mixtures. If someone asks "Can cars use
    methanol for fuel?" an honest person must say one of:
    - "No."
    - "In theory yes, but in practice no"
    - "That depends. Why do you ask? Is your question purely theoretical
    or are you contemplating something like buying methanol for your
    car? By 'cars' do you mean current, mass market products or one
    of the tiny fraction of a tenth of a percent that are built to
    use methanol?"
    Anyone who says simply "Yes" today is a dangerous fraud of a crook or
    a dangerous Cliff Claven of a poseur.


    >However, you did a good job of drawing the cases where bridging would
    >fail. And you could also have drawn the cases where it would work
    >without a hitch. But again, these are implementation details. They DO
    >NOT change the answers to how frames would be routed among different
    >topologies, for example, which was a specific question asked by the OP.


    On the contrary, the correct OSI 7 Layer Model answer that ignores
    implementation details is "No, you must use a higher layer gateway."


    >Rather than answer that question, you prefer to imply to the OP that "it
    >can't be done."


    Given the choice between only "You can bridge FDDI and Ethernet" and
    "You cannot bridge FDDI and Ethernet," an honest person must answer "It
    can't be done." Almost no one cares about purely academic layering
    theorizing about whether you can bridge FDDI and Etheret. Even those
    who do care about the pure theory need to be told "Yes if the media are
    sufficiently compatible" instead of the false and misleading "Yes."

    I'm sure the designers of those fraudulent FDDI-Ethernet bridges meant
    well and that what happened was that they were told by a guru that knew
    all there is to know about the 7 Layer Model and related issues not to
    worry and just go ahead.

    Essentially everyong asks the question for specific reasons such as:
    1. apprentice network engineers care about what they might encounter
    or need to build.
    2. sales people for bridges care, so that they can avoid selling things
    that will not work in particular installations.
    2. equipment designers need to know what design and build.
    3. salesslime care so that they can tune their lies to fit their frauds
    4. poseurs who like to hear the sounds of their own voices.
    In all of those cases except #3 and #4, the least wrong short answer is
    "It cannot be done."

    This a particular case of the evils of ISO 7 Layer gurus. Most people
    who ask layering questions have specific problems that they hope to
    understand and solve. They need to know how to make their boxes talk
    to each other or whether designing a dumb, transparent FDDI-Ethernet
    bridge is a good idea. Only the as yet unscared and the Layer Model
    gurus don't know that their purely theoretical, usually wrong in practice
    used bull food is toxic. The more common flavor of 7 Layer Guru used
    bull food is the much less harmful "No, X can't be done because that
    would violate layering."


    Vernon Schryver vjs@rhyolite.com

  9. Re: Token ring and ethernet on same physical network?

    Vernon Schryver wrote:

    > Essentially everyong asks the question for specific reasons such as:
    > 1. apprentice network engineers care about what they might encounter
    > or need to build.
    > 2. sales people for bridges care, so that they can avoid selling things
    > that will not work in particular installations.
    > 2. equipment designers need to know what design and build.
    > 3. salesslime care so that they can tune their lies to fit their frauds
    > 4. poseurs who like to hear the sounds of their own voices.
    > In all of those cases except #3 and #4, the least wrong short answer is
    > "It cannot be done."


    Check back at the top of the thread. Kaja, the OP, appears to be a
    student. He is asking many questions, in several threads, about what
    an intructor or professor is saying, or about what he is reading in
    books. With the exception of equipment vendors, who have some of the
    best experts in the field, all the others you listed are implementors
    of existing technology. Not innovators. So they don't need to
    understand what is at the base of all of this. They can go along fine
    with rules of thumb.

    Here is what his question was:

    -----------start quote----------------
    "subnetting allows a single physical network to be logically
    divided regardless of the physical layout of a network, since
    it is possible to divide a physical network into several
    subnets by configuring different host computers to use
    different routers."

    I can see how that is possible if network ( which we
    logically divided into two subnetworks S1 and S2) has star
    topology, since HUB can send the incoming packet
    intended for S1 through the ports which connect to the
    computers residing on S1.

    But I don't understand how we can, with the help of
    subnetting, logically divide Token Ring and Ethernet
    network. For starters, how can Ethernet and Token ring
    exists together as one physical network? Perhaps there
    is a device similar to HUB which uses some ports to
    connect Token ring computers and other ports connect
    Ethernet computers --> the device then based on subnet
    ID decides through which ports will the incoming packet
    go?! But other than that, I don't understand how else it
    would eb possible!
    -----------------end quote-------------------

    I'm sorry if it offends you, but a response of "it can't be done" is
    entirely wrong in this context. And furthermore, foaming at the mouth
    is not helpful when answering such questions. People have a right to
    ask questions and not be yelled at or ridiculed.

    Bert


  10. Re: Token ring and ethernet on same physical network?

    In article <1183234085.245334.237140@q69g2000hsb.googlegroups. com>,
    Albert Manfredi wrote:

    >> Essentially everyong asks the question for specific reasons such as:
    >> 1. apprentice network engineers care about what they might encounter
    >> or need to build.


    >Check back at the top of the thread. Kaja, the OP, appears to be a
    >student.


    Yes, an "apprentice network engineer" who cares (or should care)
    about what he might encounter or need to build in the real world.
    He should be told either the short answer "Bridging disparate media
    can't be done," the medium answer "If you do it, you run significant
    risks of major problems," or a long answer detailing some of the
    ways things break.


    > He is asking many questions, in several threads, about what
    >an intructor or professor is saying, or about what he is reading in
    >books.


    I'm not sure if it was a couple of his reports of entirely wrong
    words from instructors that reminded me that too many instructors
    are perfectly qualified to write the more wrong than right toxic
    waste that dilutes the ads in the trade rags. Then there were the
    alleged quotes from books about subnetting that were never right
    and are just silly today with CIDR.


    > With the exception of equipment vendors, who have some of the
    >best experts in the field, all the others you listed are implementors
    >of existing technology. Not innovators. So they don't need to
    >understand what is at the base of all of this. They can go along fine
    >with rules of thumb.


    The right rule of thumb is "Never try to bridge differing kinds of media."


    >Here is what his question was:
    >
    >-----------start quote----------------
    >"subnetting allows a single physical network to be logically
    >divided regardless of the physical layout of a network, since
    >it is possible to divide a physical network into several
    >subnets by configuring different host computers to use
    >different routers."
    >
    >I can see how that is possible if network ( which we
    >logically divided into two subnetworks S1 and S2) has star
    >topology, since HUB can send the incoming packet
    >intended for S1 through the ports which connect to the
    >computers residing on S1.
    >
    >But I don't understand how we can, with the help of
    >subnetting, logically divide Token Ring and Ethernet
    >network. For starters, how can Ethernet and Token ring
    >exists together as one physical network? Perhaps there
    >is a device similar to HUB which uses some ports to
    >connect Token ring computers and other ports connect
    >Ethernet computers --> the device then based on subnet
    >ID decides through which ports will the incoming packet
    >go?! But other than that, I don't understand how else it
    >would eb possible!


    The right answer for that question, particularly for an uninformed
    student is clearly
    "The device similar to a HUB that you are looking for is an IP router."
    If you want to give a longer answer, you might add
    "Many other tactics are sometimes possible, ranging from translucent
    (not transparent) bridging to application layer gateways, but all
    are likely to be less satisfactory than an IP router and break
    applications."

    I think the student needed to hear about encapsulation, RFC 2470, and
    RFC 894. Talk about bridging was misplaced and confusing, even if it
    had been accurate (included caveats about translucence). He probably
    also needs to hear about IP routing, but not yet Spanning Tree.


    The implicit claim that it is generally possible to bridge all kinds
    of media is simply wrong. When you do try to bridge between differing
    kinds of media, you _always_ lose something, if it works at all.
    If all media had the same features and functions, then there would
    be only one kind of media.

    How do you bridge between Ethernet and SLIP? If no local applicatios
    care about whatever is lost in the bridge, you will get away with
    it. If someone has an application that cares, you will be the
    target of curses. For example, bridging between Token Ring and
    FDDI loses the MTU hints of TR source routes and FDDI priorities
    and bandwidth allocation.

    Yes, you can gateway IP packets among all kinds of link layer media,
    but the OSI 7 Layer Model says that the way to do that is with an
    IP router.


    Vernon Schryver vjs@rhyolite.com

  11. Re: Token ring and ethernet on same physical network?

    In article ,
    Vernon Schryver wrote:

    >When you do try to bridge between differing
    >kinds of media, you _always_ lose something, if it works at all.
    >If all media had the same features and functions, then there would
    >be only one kind of media.


    I'm not clear on how your statement fits in with the many
    fibre/copper transceivers in existance? I would normally think
    of "100 Base-TX" and "100 Base-SX" as being different kinds of
    media, but very few devices do switching or routing at the
    optical level without ducking into and back out of copper.


  12. Re: Token ring and ethernet on same physical network?

    In article ,
    Walter Roberson wrote:

    >>When you do try to bridge between differing
    >>kinds of media, you _always_ lose something, if it works at all.
    >>If all media had the same features and functions, then there would
    >>be only one kind of media.

    >
    >I'm not clear on how your statement fits in with the many
    >fibre/copper transceivers in existance? I would normally think
    >of "100 Base-TX" and "100 Base-SX" as being different kinds of
    >media, but very few devices do switching or routing at the
    >optical level without ducking into and back out of copper.


    Let's not distinguish copper wire from every other because you can
    never get the diameters exactly the same. Or claim that two otherwise
    identical link layers differ in kind depending on connectors (e.g.
    the many flavors of fiber connector). We are talking about network
    addresses, hubs, routers, bridges, packets, the link layer, and upper
    layers. In this context 100BASE-TX and 100BASE-SX are identical.

    Sometimes bridging works, but other times you need a routers. Sometimes
    a router is not enough and you need an application layer gateway.
    If you can bridge two sets of wires or fibers without losing anything
    visible in the link layer and above, then they are the same kind of
    media. Yes, that's the tautology of "if bridging works, then bridging
    works." However, "works" is objective instead of driven by sales quotas
    and cloud charting engineering of the kind that gateways email between
    SMTP and x.400 with a router that is supposed to translate between IP
    and CLNP.


    Vernon Schryver vjs@rhyolite.com

+ Reply to Thread