RE: Setting the appropriate MTU

From: Davenport, David M (CRD) ([email protected])
Date: Tue Feb 15 2000 - 15:18:22 EST

  • Next message: Julien Godard: "RE: Setting the appropriate MTU"

    Now you've added mobility to the equation...

    With Inmarsat-M/Mini-M and other low data-rate, mobile satellite services
    an appropriate MTU selection is driven primarily by the Rician/Rayleigh
    fading channel and the trade-off:

    Large data frames to increase throughput efficiency
                                  - vs. -
    Small data frames to reduce required retransmission due to fading/shadowing

    For fixed site satellite with BER < 10^(-5), I'd suggest IP fragmentation along the
    entire route dominates MTU selection...and the bigger the better.

    David M. Davenport
    GE Corporate Research and Development
    One Research Circle, M/S KWC421
    Niskayuna, NY 12309
    Phone/Fax: (518) 387-5041/4042

    ***** Opinions contained herein are my own
         and not necessarily those of my employer *****

    -----Original Message-----
    From: Johnson, Gregory LCDR [mailto:[email protected]]
    Sent: Tuesday, February 15, 2000 10:39 AM
    To: Julien Godard; 'Testasecca, Mariano'
    Cc: '[email protected]'
    Subject: RE: Setting the appropriate MTU

    Bigger is not better for slow networks. For a link speed of 4kbps you will
    need to use much smaller MTU sizes since the transmission time for a packet
    gets too long otherwise. It will also depend upon what the latency is in the
    link because the total transmission time for a packet is a function of the
    size and delay. For a typical GEO link, I would expect that you have
    anywhere up to 600ms depending upon what kind of FEC and interleaving is
    done on the link. I have found for Inmarsat-M/Mini-M type links (2.4 kbps)
    that an MTU size of 500 works well. You also need to adjust the initial RTO
    value larger (the default is 3 secs, try 5 secs). If you don't do this, you
    will suffer a lot of timeouts and retransmissions. I'm not that familiar
    with the protocol Citrix uses, but with only 4kbps of bandwidth, you also
    don't want to run multiple simultaneous connections across your link as one
    connection will probably saturate the link.
     

    -Greg

    LCDR Gregory W. Johnson
    Ass't Prof. Electrical Engineering

    USCG Academy
    New London CT

    860-444-8683

    -----Original Message-----
    From: Julien Godard [mailto:[email protected]]
    Sent: Tuesday, February 15, 2000 4:59 AM
    To: 'Testasecca, Mariano'
    Cc: '[email protected]'
    Subject: RE: Setting the appropiate MTU

    for performance, the bigger the better !
    But it may depend also on your network because you may want to avoid
    fragmentation.
    You can test your MTU with some free tools such as easyMTU
    http://members.tripod.com/~EasyMTU/ <http://members.tripod.com/~EasyMTU/>
    For example if you use ethernet link, the MTU is 1500, leading to 1448 bytes
    of data in each segment (20 IP + 20 TCP + 12 TCP timestamp option + 1448
    data). If your router support this size without fragmentation, it's fine !

            -----Original Message-----
    From: Testasecca, Mariano [SMTP:[email protected]]
    Sent: Monday, February 14, 2000 7:49 PM
    To: '[email protected]'
    Subject: Setting the appropiate MTU

            What MTU value would be appropiate for Windows 95/98 terminals
    working as a
    Windows NT 4.0 running Citrix MetaFrame under
    a satellite link of a VSAT that provides 4kbps of bandwidth?

            Thanks to all.

            Mariano Testasecca
    SIEMENS S.A.
    Division Electromedicina
    Servicios de Salud - SHS
    Bolivar 177 1o Piso
    *+54-11-4340-8400 int. 2796
    * mailto:[email protected]
    <mailto:[email protected]>



    This archive was generated by hypermail 2b29 : Tue Feb 15 2000 - 16:37:08 EST