Packet switching and ATM - TCP-IP

This is a discussion on Packet switching and ATM - TCP-IP ; hello Doesn't packet switching refer to routing a packet to its final destination? If so, then packet switching happens at layer 3 ( network layer ) of OSI model. So why do we call ATM fast packet switching protocol, if ...

+ Reply to Thread
Results 1 to 11 of 11

Thread: Packet switching and ATM

  1. Packet switching and ATM

    hello

    Doesn't packet switching refer to routing a packet to its final
    destination? If so, then packet switching happens at layer 3 ( network
    layer ) of OSI model.

    So why do we call ATM fast packet switching protocol, if ATM resides
    at layer 2 ( data link layer ) of OSI model, and as such it has
    nothing to do with packet switching which, as I've said before,
    happens at layer 3?!


    thank you


  2. Re: Packet switching and ATM

    Actually, I'm also a bit confused as to why we use word a SWITCHING
    ( packet switching ) to describe a process of directing a packet to
    its final destination?



  3. Re: Packet switching and ATM

    On 24 Jun, 11:57, kaja_love...@yahoo.com wrote:
    > Actually, I'm also a bit confused as to why we use word a SWITCHING
    > ( packet switching ) to describe a process of directing a packet to
    > its final destination?


    When an ATM virtual circuit is made, each ATM switch will direct the
    ATM cells (similar to IP packets) to the next switch based on their
    VPI and VCI pairs. The next switch is determined from a dumb lookup
    on these 3-4 bytes. In ATM all routing is performed up front when the
    virtual circuit is established.

    Switching is conceptually simpler than routing. A 'switch' operation
    is a simple lookup from input port to output port on the ATM switch,
    an IP 'route' operation is more complex.


  4. Re: Packet switching and ATM

    wrote in message
    news:1182681334.113810.159860@n60g2000hse.googlegr oups.com...
    > hello
    >
    > Doesn't packet switching refer to routing a packet to its final
    > destination?


    sometimes. Switching usually implies hardware based forwarding, so could be
    bridging (ie L2), or L3.

    Some boxes now providing "hardware switching" for higher level protocols -
    for example a hardware accelerated firewall supporting NAT.

    The terms switching and routing are not used consistently.

    If so, then packet switching happens at layer 3 ( network
    > layer ) of OSI model.


    Sort of. If you want unambiguous statements, then define your terms
    accurately - and switching has too many inconsistent meanings.
    >
    > So why do we call ATM fast packet switching protocol, if ATM resides
    > at layer 2 ( data link layer ) of OSI model, and as such it has
    > nothing to do with packet switching which, as I've said before,
    > happens at layer 3?!


    because you are mixing up the OSI model with real life, and real protocols
    do not map exactly into the model.

    You need to remember that the whole reason for a model is to abstract away
    detail...

    And FWIW ATM is usually described as cell switching.

    But since ATM is connection oriented, you can argue a VC is not strictly a
    layer 2 service.

    >
    >
    > thank you
    >

    --
    Regards

    stephen_hope@xyzworld.com - replace xyz with ntl



  5. Re: Packet switching and ATM

    "tomalt" wrote in message
    news:1182697096.964374.57900@p77g2000hsh.googlegro ups.com...
    > On 24 Jun, 11:57, kaja_love...@yahoo.com wrote:
    > > Actually, I'm also a bit confused as to why we use word a SWITCHING
    > > ( packet switching ) to describe a process of directing a packet to
    > > its final destination?

    >
    > When an ATM virtual circuit is made, each ATM switch will direct the
    > ATM cells (similar to IP packets) to the next switch based on their
    > VPI and VCI pairs. The next switch is determined from a dumb lookup
    > on these 3-4 bytes. In ATM all routing is performed up front when the
    > virtual circuit is established.
    >
    > Switching is conceptually simpler than routing. A 'switch' operation
    > is a simple lookup from input port to output port on the ATM switch,
    > an IP 'route' operation is more complex.


    That may have been true several years ago, but modern L3 IP switches do IP
    forwarding between subnets in hardware. So, your statement has been
    overtaken by improved hardware design and higher chip complexity......

    Although L3 forwarding might be a bit more complex than cell switching the
    same functions apply - lookup, decide where to send, adjust some fields in
    the data, send, and it can be implemented in hardware.
    >

    --
    Regards

    stephen_hope@xyzworld.com - replace xyz with ntl



  6. Re: Packet switching and ATM

    Bonjour Kaja,

    > Actually, I'm also a bit confused as to why we use word a SWITCHING
    > ( packet switching ) to describe a process of directing a packet to
    > its final destination?


    Today, the switching function is technically similar to the routing
    function. We reserve the word switching for the transmission of the
    layer 2 protocols, and the word routing for the transmission of the
    layer 3 protocols.

    Sometimes we used the word routing in a generic sense, refering to
    both layer 2 and layer 3 protocols.

    > Doesn't packet switching refer to routing a packet to its final
    > destination?


    The word routing is ambiguous here, and used in the sense of
    transmission, don't use it in your statement. You can say that
    switching refer to transmit a packet.

    > So why do we call ATM fast packet switching protocol


    ATM protocol is not fast, you can transmit cells (which are packets of
    53 bytes) at the rate of 8 kbit/s or 40 Gbit/s. The rate of the ATM
    protocol is those of the layer 1 protocol.

    Today, ATM only refers to the layer 2 protocol. At the beginning ATM
    meant either ATM architecture (or ATM model), including all the stack
    of protocols, either the layer 2 protocol. So it can be a source of
    confusion.

    Best regards,
    Michelot


  7. Re: Packet switching and ATM

    wrote:

    > Doesn't packet switching refer to routing a packet to its final
    > destination? If so, then packet switching happens at layer 3 ( network
    > layer ) of OSI model.


    I will give you my own answers to your questions, although I agree for
    the most part with what the other responders said. It's mostly a
    question of nuance and interpretation.

    When you speak of "packet," in certain IP-centric circles, you are
    referring to TCP/IP or UDP/IP packets. The Ethernet frames or ATM cells
    that carry these IP packets are not called "packets" at all!

    So, the switching done on ATM cells or Ethernet frames does not
    necessarily carry the payload all the way to its destination. It will
    more typically carry this IP payload to an IP router, which then might
    put the stuff on a totally different type of network.

    > So why do we call ATM fast packet switching protocol, if ATM resides
    > at layer 2 ( data link layer ) of OSI model, and as such it has
    > nothing to do with packet switching which, as I've said before,
    > happens at layer 3?!


    In an IP-centric discussion, ATM is *not* called "fast packet
    switching," but rather "(perhaps fast) cell switching." Yes, it might be
    true that today's routers may also route packets in hardware, but that's
    a digression. The terminology is what is confusing you, and that
    terminology can be perfectly consistent. We're not debating the reasons
    why ATM was invented in the first place, after all.

    Similarly, Ethernet switches do "frame switching," not "packet
    switching." Exceptions are those ATM switches and Ethernet switches that
    are also IP routers. That type of box will indeed also perform a routing
    function (packet switching).

    IMO, the ISO/OSI model works fine here, and certainly can help organize
    your thinking.

    In short, think of "packet" as being something at layer 3, and "frame"
    or "cell" as something at layer 2. And just accept that many people
    won't get too religious aboiut such terminology.

    Bert


  8. Re: Packet switching and ATM

    In article <1182681334.113810.159860@n60g2000hse.googlegroups. com>,
    kaja_love160@yahoo.com wrote:

    > hello
    >
    > Doesn't packet switching refer to routing a packet to its final
    > destination? If so, then packet switching happens at layer 3 ( network
    > layer ) of OSI model.
    >
    > So why do we call ATM fast packet switching protocol, if ATM resides
    > at layer 2 ( data link layer ) of OSI model, and as such it has
    > nothing to do with packet switching which, as I've said before,
    > happens at layer 3?!


    "Packet switching" doesn't refer to any specific layer or networking
    model. In my experience, it's just a general term that means that each
    packet is treated independently by the networking components. The
    intent is to distinguish it from "circuit switching", where prior to
    sending any data it's necessary to allocate all the network resources
    for that particular communication channel.

    --
    Barry Margolin, barmar@alum.mit.edu
    Arlington, MA
    *** PLEASE post questions in newsgroups, not directly to me ***
    *** PLEASE don't copy me on replies, I'll read them in the group ***

  9. Re: Packet switching and ATM

    Bonjour Barry,

    > "Packet switching" doesn't refer to any specific layer or networking
    > model. In my experience, it's just a general term that means that each
    > packet is treated independently by the networking components. The
    > intent is to distinguish it from "circuit switching", where prior to
    > sending any data it's necessary to allocate all the network resources
    > for that particular communication channel.


    Take care, you have packet switching is connectionless mode, and
    packet swiching in connection mode where you have also to allocate all
    the network resources previously to send the user data.

    Best regards,
    Michelot


  10. Re: Packet switching and ATM

    thank you all for your kind help

    cheers



  11. Re: Packet switching and ATM

    kaja_love160@yahoo.com wrote:

    > Doesn't packet switching refer to routing a packet to its final
    > destination? If so, then packet switching happens at layer 3 ( network
    > layer ) of OSI model.


    No. It refers to the nature of the network itself, as opposed to a
    virtual circuit for example.

+ Reply to Thread