SORT BY:

LIST ORDER
THREAD
AUTHOR
SUBJECT


SEARCH

IPS HOME


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

    Re: Data in SCSI Response or SCSI Data



    
    
    Is there anything preventing your hypothetical hardware implementor
    to send always good status within the last block of data?
    
    Does the hardware implementor make more decisions when finding status
    allone in a response block or within the header of the last block?
    
    Software based initiators will be, IMHO, a very important part of the
    market -
    to have them go through two calls is not wise.
    
    On the other hand I am not religious about it.
    
    However - sending "good status"
    with the last piece of data has been part of the the I/O scene for a long
    time
    and I did attempt to keep the status quo.
    
    Julo
    
    
    Stephen Bailey <steph@cs.uchicago.edu> on 25/08/2000 19:14:48
    
    Please respond to Stephen Bailey <steph@cs.uchicago.edu>
    
    To:   ips@ece.cmu.edu
    cc:    (bcc: Julian Satran/Haifa/IBM)
    Subject:  Re: Data in SCSI Response or SCSI Data
    
    
    
    
    > Again I have to agree.  Not everyone writing this protocol is developing
    a
    > hardware device.  Every time I (or any hardware vendor) sends one less IP
    > packet down the wire is one less chance of it getting lost.
    
    I do not think concessions should be made to software implementations
    at the expense of hardware implementations.  Our experience is nobody
    cares about a software network storage implementation if there are not
    also hardware implementations.  The value proposition completely dries
    up.  With available hardware implementations software implementations
    become interesting.
    
    I agree with Y.P. Cheng that implementing hardware for three ways to
    find status IS a going to be a substantial irritation.  More
    importantly, the outbound path of a hardware implementation (which
    will include most targets) is very likely to chose to implement the
    single easiest way for it, and use it in all cases.  In other words,
    if data in one packet with RDMA information and status in a second
    packet without is the natural way (seems like it), that's what you're
    going to get coming back to the initiator.  Your software `fast path'
    will never get exercised.
    
    Even host `software' implementations are going to be able to support
    various levels of hardware acceleration.  In the case of using a
    hardware accelerated adapter, the host software will not see the data
    packet, only the status.
    
    Steph
    
    
    
    


Home

Last updated: Tue Sep 04 01:07:44 2001
6315 messages in chronological order