Maybe the wording on this question is sloppy. - TCP-IP

This is a discussion on Maybe the wording on this question is sloppy. - TCP-IP ; I tried posting this question in a few other forums and got no response. I'm suspecting that perhaps the question off topic. The following question stems from the bottom of page 55 in the book "Unix Network Programming: The Sockets ...

+ Reply to Thread
Results 1 to 4 of 4

Thread: Maybe the wording on this question is sloppy.

  1. Maybe the wording on this question is sloppy.

    I tried posting this question in a few other forums and got no
    response. I'm suspecting that perhaps
    the question off topic.

    The following question stems from the bottom of page 55 in the book
    "Unix Network Programming: The Sockets Networking API" by Stevens,
    Fenner, and Rudoff.

    "Older SLIP links often used an MTU of 1006 or 296 bytes"

    I'm assuming the 296 comes from 2^8 + 20 bytes for the TCP header + 20
    bytes for the IP header. What does the 1006 represent in this case?

    "The minimum link MTU for IPv4 is 68 bytes."

    20 bytes is for the fixed header and 8 bytes is for the fragment
    offset. What about other 40 bytes? I did some googling and found the
    following

    http://en.wikipedia.org/wiki/MTU_(networking)

    Near the end of the article, they have the following:
    4 byte FCS (Ethernet Frame Check Sequence, not always present)
    14 byte Ethernet header
    8 byte SAR (Segmentation and Reassembly) AAL5 (ATM Adaptation Layer
    5),
    for Ethernet encapsulation in ATM
    8 byte PPPoE (Point-to-Point Protocol over Ethernet) header
    10 byte RFC-1483 LLC/SNAP (Logical Link Control/Subnetwork Access
    Protocol) encapsulation

    If I don't account for the FCS, this adds up the other 40 bytes.

    Would these be the options in this case?



    Chad


  2. Re: Maybe the wording on this question is sloppy.

    In article <1165079543.298435.205720@l12g2000cwl.googlegroups. com>,
    "grocery_stocker" wrote:

    > I tried posting this question in a few other forums and got no
    > response. I'm suspecting that perhaps
    > the question off topic.


    It's on topic. It's just not clear what problem you're trying to solve.
    If you're just curious about the history, there's a good chance no one
    really remembers the specific reasons.

    >
    > The following question stems from the bottom of page 55 in the book
    > "Unix Network Programming: The Sockets Networking API" by Stevens,
    > Fenner, and Rudoff.
    >
    > "Older SLIP links often used an MTU of 1006 or 296 bytes"
    >
    > I'm assuming the 296 comes from 2^8 + 20 bytes for the TCP header + 20
    > bytes for the IP header. What does the 1006 represent in this case?
    >
    > "The minimum link MTU for IPv4 is 68 bytes."
    >
    > 20 bytes is for the fixed header and 8 bytes is for the fragment
    > offset. What about other 40 bytes? I did some googling and found the
    > following


    The fragment offset is part of the fixed IP header. Are you talking
    about the option bytes?

    >
    > http://en.wikipedia.org/wiki/MTU_(networking)
    >
    > Near the end of the article, they have the following:
    > 4 byte FCS (Ethernet Frame Check Sequence, not always present)
    > 14 byte Ethernet header
    > 8 byte SAR (Segmentation and Reassembly) AAL5 (ATM Adaptation Layer
    > 5),
    > for Ethernet encapsulation in ATM
    > 8 byte PPPoE (Point-to-Point Protocol over Ethernet) header
    > 10 byte RFC-1483 LLC/SNAP (Logical Link Control/Subnetwork Access
    > Protocol) encapsulation
    >
    > If I don't account for the FCS, this adds up the other 40 bytes.
    >
    > Would these be the options in this case?


    Not likely, since ATM, PPP, PPPoE, and LLC didn't exist when the IP
    specification was written. I think the 40 bytes is to allow for a
    reasonable amount of payload, e.g. 20 bytes of TCP header and some
    application layer data. The whole point of networking is to transfer
    "information", and if you allow the MTU to be too small you'll use it
    all up just sending headers!

    --
    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 ***

  3. Re: Maybe the wording on this question is sloppy.

    "grocery_stocker" wrote:
    >I tried posting this question in a few other forums and got no
    >response. I'm suspecting that perhaps
    >the question off topic.
    >
    >The following question stems from the bottom of page 55 in the book
    > "Unix Network Programming: The Sockets Networking API" by Stevens,
    > Fenner, and Rudoff.
    >
    >"Older SLIP links often used an MTU of 1006 or 296 bytes"
    >
    >I'm assuming the 296 comes from 2^8 + 20 bytes for the TCP header + 20
    > bytes for the IP header. What does the 1006 represent in this case?


    I don't remember the answer to this question; it has been many years
    since I've done any work with SLIP. A few searches turned up the
    following text in RFC 1055:

    Because there is no 'standard' SLIP specification, there is no real
    defined maximum packet size for SLIP. It is probably best to
    accept the maximum packet size used by the Berkeley UNIX SLIP
    drivers: 1006 bytes including the IP and transport protocol headers
    (not including the framing characters). Therefore any new SLIP
    implementations should be prepared to accept 1006 byte datagrams
    and should not send more than 1006 bytes in a datagram.

    There is also a table in RFC 1191 that indicates an MTU of 1006 was
    used for BBN IMP (1822) interfaces.

    I don't remember much about VAX or Sun kernels either; perhaps 1006 +
    some overhead was necessary in order for packets to optimally fit into
    pages.

    --gregbo
    gds at best dot com

  4. Re: Maybe the wording on this question is sloppy.

    On Sat, 02 Dec 2006, in the Usenet newsgroup comp.protocols.tcp-ip, in article
    , Barry Margolin wrote:

    >In article <1165079543.298435.205720@l12g2000cwl.googlegroups. com>,
    > "grocery_stocker" wrote:
    >
    >> I tried posting this question in a few other forums and got no
    >> response.


    You should allow adequate time for the article to propagate, for others to
    see, and respond, and their response to propagate back to you. I answered
    your 'Sat, 2 Dec 2006 01:35:12 +0000' post in comp.os.linux.networking
    on 'Sat, 02 Dec 2006 12:44:22 -0600' which is about 18 hours later.

    >If you're just curious about the history, there's a good chance no one
    >really remembers the specific reasons.


    news://alt.folklore.computers ;-)

    >> I'm assuming the 296 comes from 2^8 + 20 bytes for the TCP header + 20
    >> bytes for the IP header. What does the 1006 represent in this case?


    296 = Stevens 'TCP/IP Illustrated Volume 1' Section 2.10 - interactive
    delay on a 9600 BPS link. The 1006 figure seems to have been for a 33.6
    modem (or, I suppose, a 38400 BPS serial link). Of course today, there is
    little reason to mess with the MTU as the defaults will usually work fine
    (exception - some mis-configured PPP-over-mumble abortions).

    >> "The minimum link MTU for IPv4 is 68 bytes."
    >>
    >> 20 bytes is for the fixed header and 8 bytes is for the fragment
    >> offset. What about other 40 bytes? I did some googling and found the
    >> following

    >
    >The fragment offset is part of the fixed IP header. Are you talking
    >about the option bytes?


    RFC0791 page 25 fourth paragraph

    >> Would these be the options in this case?

    >
    >Not likely, since ATM, PPP, PPPoE, and LLC didn't exist when the IP
    >specification was written. I think the 40 bytes is to allow for a
    >reasonable amount of payload, e.g. 20 bytes of TCP header and some
    >application layer data.


    RFC0791 page 2 (and 15 et.seq.). RFC0793 also includes options.

    >The whole point of networking is to transfer "information", and if you
    >allow the MTU to be too small you'll use it all up just sending headers!


    Efficiency? Wazzat?

    Old guy


+ Reply to Thread