Does ARP Belong to Layer 2 Or Layer 3 OSI Reference Model??? - TCP-IP

This is a discussion on Does ARP Belong to Layer 2 Or Layer 3 OSI Reference Model??? - TCP-IP ; Rick Jones wrote: > David Schwartz wrote: >> ARP translates layer 3 addresses into layer 2 addresses. As such, it >> is part of the implementation of layer 3 using layer 2. I don't see >> any of those links ...

+ Reply to Thread
Page 2 of 2 FirstFirst 1 2
Results 21 to 26 of 26

Thread: Does ARP Belong to Layer 2 Or Layer 3 OSI Reference Model???

  1. Re: Does ARP Belong to Layer 2 Or Layer 3 OSI Reference Model???

    Rick Jones wrote:
    > David Schwartz wrote:
    >> ARP translates layer 3 addresses into layer 2 addresses. As such, it
    >> is part of the implementation of layer 3 using layer 2. I don't see
    >> any of those links claiming ARP is a layer 2 protocol, in fact, they
    >> claim the reverse.

    >
    >> The layer 2 protocol will work just fine without ARP.

    >
    > Layer 3 protocols can also work just fine without ARP. It might not
    > be as convient, but it can work. ARP also isn't the only thing out
    > there ever to translate IP to MAC addresses.


    Strictly speaking there doesn't need to be any mapping at all since one
    could use broadcast mode on all Ethernet frames and let the IP layer
    discard the datagrams not intended for that router or computer.

  2. Re: Does ARP Belong to Layer 2 Or Layer 3 OSI Reference Model???

    Jim Logajan a écrit :
    >
    > Strictly speaking there doesn't need to be any mapping at all since one
    > could use broadcast mode on all Ethernet frames and let the IP layer
    > discard the datagrams not intended for that router or computer.


    Doing so would break IP communications and routing.

    RFC 1122 (requirements for IPv4 hosts) says :

    When a host sends a datagram to a link-layer broadcast address,
    the IP destination address MUST be a legal IP broadcast or IP
    multicast address.

    A host SHOULD silently discard a datagram that is received via
    a link-layer broadcast (see Section 2.4) but does not specify
    an IP multicast or broadcast destination address.

    RFC 1812 (requirements for IPv4 routers) says :

    A router MUST NOT forward any packet that the router received as a
    Link Layer broadcast, unless it is directed to an IP Multicast
    address.

  3. Re: Does ARP Belong to Layer 2 Or Layer 3 OSI Reference Model???

    On Feb 6, 2:23*pm, Pascal Hambourg
    wrote:
    > Jim Logajan a écrit :
    >
    > > Strictly speaking there doesn't need to be any mapping at all since one
    > > could use broadcast mode on all Ethernet frames and let the IP layer
    > > discard the datagrams not intended for that router or computer.

    >
    > Doing so would break IP communications and routing.
    >
    > RFC 1122 (requirements for IPv4 hosts) says :
    >
    > * * When a host sends a datagram to a link-layer broadcast address,
    > * * the IP destination address MUST be a legal IP broadcast or IP
    > * * multicast address.
    >
    > * * A host SHOULD silently discard a datagram that is received via
    > * * a link-layer broadcast (see Section 2.4) but does not specify
    > * * an IP multicast or broadcast destination address.
    >
    > RFC 1812 (requirements for IPv4 routers) says :
    >
    > * * A router MUST NOT forward any packet that the router received as a
    > * * Link Layer broadcast, unless it is directed to an IP Multicast
    > * * address.


    Yes, exactly. That's why I think a better way of settling this is to
    say that ARP operates entirely at L2 only, even though the service it
    provides is a service *for* L3, as David points out. The addressing of
    ARP frames, I mean the addresses used to deliver the frames, are only
    L2 addresses. The same applies to RARP.

    Saying this another way, even if ARP provides a service *for* L3, this
    does not mean that it is providing L3 services. Or that it uses L3
    services. It does not, and I think we can all agree on that.

    Also, ARP is not exclusive to IP, and IP can and does work well
    without ARP, depending on the underlying network. And other L3
    protocols, such as DECnet, never used ARP (that I'm aware of).

    Bert

  4. Re: Does ARP Belong to Layer 2 Or Layer 3 OSI Reference Model???

    Pascal Hambourg wrote:
    > Jim Logajan a écrit :
    >>
    >> Strictly speaking there doesn't need to be any mapping at all since one
    >> could use broadcast mode on all Ethernet frames and let the IP layer
    >> discard the datagrams not intended for that router or computer.

    >
    > Doing so would break IP communications and routing.
    >
    > RFC 1122 (requirements for IPv4 hosts) says :
    >
    > When a host sends a datagram to a link-layer broadcast address,
    > the IP destination address MUST be a legal IP broadcast or IP
    > multicast address.
    >
    > A host SHOULD silently discard a datagram that is received via
    > a link-layer broadcast (see Section 2.4) but does not specify
    > an IP multicast or broadcast destination address.
    >
    > RFC 1812 (requirements for IPv4 routers) says :
    >
    > A router MUST NOT forward any packet that the router received as a
    > Link Layer broadcast, unless it is directed to an IP Multicast
    > address.


    All true and acceded but when I wrote "strictly speaking" my intended
    meaning was that _theoretically_ all the information that IP needs to do
    its job _could_ be done without access to any header or trailer information
    from the link layer.

    Anyway, the RFC 1812 quoted rule applies "only to Link Layer protocols that
    allow broadcasts to be distinguished." All I have to do is define a "Jim's
    Ethernet version 0.5" that only ever uses MAC broadcast addresses to dance
    past those requirements. Of course my "Jim's Ethernet version 0.5" isn't
    compatible with Ethernet past the physical layer....

  5. Re: Does ARP Belong to Layer 2 Or Layer 3 OSI Reference Model???

    On 6 Feb, 19:23, Pascal Hambourg wrote:
    > Jim Logajan a écrit :
    >
    >
    >
    > > Strictly speaking there doesn't need to be any mapping at all since one
    > > could use broadcast mode on all Ethernet frames and let the IP layer
    > > discard the datagrams not intended for that router or computer.

    >
    > Doing so would break IP communications and routing.
    >
    > RFC 1122 (requirements for IPv4 hosts) says :
    >
    > * * When a host sends a datagram to a link-layer broadcast address,
    > * * the IP destination address MUST be a legal IP broadcast or IP
    > * * multicast address.
    >
    > * * A host SHOULD silently discard a datagram that is received via
    > * * a link-layer broadcast (see Section 2.4) but does not specify
    > * * an IP multicast or broadcast destination address.
    >
    > RFC 1812 (requirements for IPv4 routers) says :
    >
    > * * A router MUST NOT forward any packet that the router received as a
    > * * Link Layer broadcast, unless it is directed to an IP Multicast
    > * * address.


    Link layer may not be OSI Data Link Layer. When I saw a book refer to
    "Link Layer" it was in the TCP/IP reference model

    And they drew it something like this (mnemonic ATIN)

    Application layer
    Transport layer
    /\
    |
    Link Layer
    |
    \/
    Internetwork Layer

    Network Interface Layer

    (wendell odom`s book, networking first step, mapped one to the other..
    Network Interface Layer is Physical Layer. Did not mention "link
    layer" I don`t think)

    The TCP/IP book published by Que ("using TCP/IP"), does show a link
    layer, in dashed lines. In the TCP/IP reference model. Not the OSI.
    (OSI has "data link layer" as mentioned)


    I think what matters is what it -does-. Function. The purpose of OSI
    reference model (a terminology of a different network architecture)
    applied to this, is just to provide a way to conceptualise it. I am
    sure their intention was to make it easier to understand.

    Routers only read layer 3, (IP). Switches only read layer 2. Hubs
    only use Layer 1. The reference model works very nicely to an
    extent. TCP has its place, Application layer protocols have their
    place. If protocols are listed at each layer, it even looks like an
    hour glass figure, and we like that
    (transport layer only has TCP and UDP, network layer typically only
    uses IP)

    The ARP protocol doesn`t really fit into one layer, it uses both Layer
    2 and Layer 3 information. That does make it rather exceptional
    compared to all or most of the other protocols and conceptual
    devices..

    And anyhow, I am not an expert, but don`t people say that the whole
    philosophy with TCP/IP , in contrast with OSI. Was that with TCP/IP
    they developed the thing and explained it afterwards in RFCs.. So RFCs
    don`t define things as definitively/encyclopedically as one would
    like.. And definitions in one RFC are only true within it. (I do not
    know if that is true for all RFCs.. I know some RFCs have higher
    status than others)

    I love definitions, but they seem to play second fiddle in networking.
    For example, a vaguely related issue - You have marketting
    departments.
    But more relevant, one can find onself eading/ searching a jungle of
    RFCs for a term, hoping for some consistency, to ascertain its
    meaning. Sometimes I wonder whether the people who came up with the
    term give a toss. Certainly they object sometimes, engineers get mad
    at marketers "..."

    And the way people use terminology is totally screwy.. I often just
    ask them what it does. And they only seem to care about that anyway.
    If the definition of a term is so elusive, then maybe the people that
    came up with it couldn`t care less!

    Or perhaps, in this case, the people that came up with it did not come
    up with it. The layers concept is from ISO for their OSI
    architecture. And it is just being applied to what was the competing
    architecture - TCP/IP. Perhaps because somebody felt TCP/IP had not
    defined things well..(though ATIN seems ok to me)

    I could be way off track here,

    I did see a post where somebody wrote of "these people" that ask
    whether ARP is layer 2 or layer 3. Nobody objected to him.
    One can either assume that they didn`t know.. Or the conclusion I
    drew..
    I got the impression that they don`t give a toss, and that they
    thought it doesn`t fit in because nobody gives a toss, and it is just
    a conceptual thing that helps to an extent and just exists to make
    things easier.
    Who is "they", well, those in the technical community, in that
    newsgroup in that thread at that time..I would think people on usenet
    would be more interested in being technically correct than people in
    real life conversations.




  6. Re: Does ARP Belong to Layer 2 Or Layer 3 OSI Reference Model???

    ARP is definetely a layer 2 protocol . With type = 0x806

    If anyone claims that ARP is on L3 , What is the protocol number for ARP ??

+ Reply to Thread
Page 2 of 2 FirstFirst 1 2