SORT BY:

LIST ORDER
THREAD
AUTHOR
SUBJECT


SEARCH

IPS HOME


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

    RE: profiles - a way to simplify iSCSI



    
    
    Bernard,
    
    I understand you well stated (and well intentioned remarks) and it is hard
    not too agree with motherhood statements.
    However I have to mention that most of the features where added with care
    and only a very few remained after selection. Peace was not what we had in
    mind - but you can't outright reject feature that have a clearer rationale
    for a class of users.
    
    However the draft addresses a wide spectrum of devices with different
    requirements (as do most of the SCSI drafts). Some can be met only with
    minimal requirement (most low-end and office raid-boxes) while others
    require higher performance/reliability/maintainability etc.
    
    Profiles could clearly cluster the space and simplify implementation and
    testing.
    
    Regards,
    Julo
    
    Bernard Aboba <aboba@internaut.com> on 25-06-2001 09:09:34
    
    Please respond to Bernard Aboba <aboba@internaut.com>
    
    To:   Julian Satran/Haifa/IBM@IBMIL
    cc:   ips@ece.cmu.edu
    Subject:  RE: profiles - a way to simplify iSCSI
    
    
    
    
    > You are still missing my point. Profiles are not proposed "in addition"
    to
    > what we have but instead of.
    > They are not meant to remove features that where introduced for a
    > legitimate reason or another but rather to limit
    > the variability in implementations and testing.
    >
    
    The best way to limit the variability in implementations and testing is
    not to include so many "features" in the first place. It's often tempting
    to resolve potential conflicts by turning out a "kitchen sink spec" but
    that's almost always a worse choice than choosing between one of the
    alternatives. Similarly, options are a losing proposition because you have
    to implement them because someone else might, but you can't guarantee
    they'll be used. So if something doesn't make it into what would have been
    the "profile" I'd question why it can't be thrown out altogether.
    
    As part of the Draft Standards application, "features" not proven to
    interoperate with independent implementations have to be removed. So if a
    "feature" was inserted merely to keep the peace, has problematic
    interoperability or is not on the slate to be independently implemented,
    we might as well toss it out.
    
    "There's never a better time to weed a garden than RIGHT NOW."
    
    
    
    
    
    
    
    


Home

Last updated: Tue Sep 04 01:04:24 2001
6315 messages in chronological order