SORT BY:

LIST ORDER
THREAD
AUTHOR
SUBJECT


SEARCH

IPS HOME


    [Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

    RE: I-D ACTION:draft-cavanna-iscsi-crc-vs-cksum-00.txt



    Julian,
    
    As there are already selectors in place for checking schemes, what reason is
    there for excluding Adler-32?  It is not code size, speed of execution nor
    difficulty in hardware implementation.  Excluding this alternative ensures
    potential loss of a competitive software solution.  You wish to build a
    bridge to carry trucks, but there is a height barrier that prevents such
    trucks.  Is a serial burst error high on the list of probable errors if CRC
    checks prevent these already?
    
    Doug
    
    > Pat,
    >
    > I did not run a check. That is very expensive.
    > With CRCs there are methods that enable you to run a check on the
    > complement code
    > (that has only 2**32 different patterns for any block length) and derive
    > from there the distances (Fujiwara has done this in 1989 for the IEEE
    > CRC-32 and about some more recent experiments I'll get back to the list).
    >
    > And that is the trouble with this whole line of argument.
    > There are no numbers to prove Adler32 or Fletcher32 and there are plenty
    > for CRCs.
    >
    > The big question is then is there anybody out there that wants to build a
    > modern bridge based only on its beauty?
    >
    > Regards,
    > Julo
    >
    >
    > pat_thaler@agilent.com on 05/03/2001 23:27:49
    >
    > Please respond to pat_thaler@agilent.com
    >
    > To:   Julian Satran/Haifa/IBM@IBMIL
    > cc:
    > Subject:  RE: I-D ACTION:draft-cavanna-iscsi-crc-vs-cksum-00.txt
    >
    >
    >
    >
    > Julian,
    >
    > I know that Hamming distance gets down to 2 for errors that are separated
    > by
    > the modulus (or a multiple of it). Is there another case?
    >
    > Pat
    >
    > > - Adler and Fletcher are weak and there is no theory behind your
    > > distribution statements, nor any simulation results as far as I
    > know.  We
    > > found that on very simple sequences the Hamming distance gets
    > > down to 2 (or
    > > lower) and the burst protection is probably not better than 16 bit.
    > There
    > > is even a simple formula for what sequences will get you false
    > codes (see
    > > bellow for a reference)
    > >
    >
    >
    >
    >
    
    


Home

Last updated: Tue Sep 04 01:05:26 2001
6315 messages in chronological order