SORT BY:

LIST ORDER
THREAD
AUTHOR
SUBJECT


SEARCH

IPS HOME


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

    RE: Framing Discussion



    
    
    I agree with David. For the sake of clarity however I would like to add
    that even for those cases
    in which buffers are anonymous (like some appliance fileservers or disk
    controllers) the data that comes from the wire are not necessarily entire
    storage data blocks and have to be placed somewhere within a storage
    buffer.
    That particular placement can be made itself anonymous only by "placing" a
    "scather-gather" burden on the
    storage end of the wire and that is either expensive or outright
    impossible.
    
    Julo
    
    Black_David@emc.com on 20/12/2000 23:40:05
    
    Please respond to Black_David@emc.com
    
    To:   ips@ece.cmu.edu
    cc:
    Subject:  RE: Framing Discussion
    
    
    
    
    > No. Why do you need a copy ? These buffers are consumed
    > by iscsi layer and may be returned later to the same
    > pool. These are just some details.
    > Why are they not interchangeable ?
    
    I have a hunch that Mohan has never worked with local filesystem
    or block device code :-).
    
    The reason is that SCSI code inside operating systems does not
    work that way.  Operating systems perform reads from SCSI devices
    by allocating memory at a layer above SCSI and then telling
    the SCSI driver to read data into the memory it has allocated.
    This is the case even when the operating system (e.g., filesystem)
    is the originator of the I/O.  The allocation operation cannot be
    moved into the SCSI driver without significant code changes to
    the OS (e.g., the device driver interface has to change, and then
    we can talk about the impact on the filesystem).  To keep
    existing code happy, the act of iSCSI "consuming" the data
    in the above scenario will include copying (or otherwise
    moving) it to the location at which it was supposed to arrive.
    
    >  This is an
    > assumption not stated anywhere.  Is there
    > a list of other assumptions that is documented ?
    
    I suggest looking at the SCSI/block device driver
    interface inside an operating system like Solaris,
    which I'm sure someone at Sun can explain :-) .
    It's very different from the interface to network devices
    (e.g., no unsolicited inbound data).
    
    --David
    
    ---------------------------------------------------
    David L. Black, Senior Technologist
    EMC Corporation, 42 South St., Hopkinton, MA  01748
    +1 (508) 435-1000 x75140     FAX: +1 (508) 497-8500
    black_david@emc.com       Mobile: +1 (978) 394-7754
    ---------------------------------------------------
    
    
    
    
    


Home

Last updated: Tue Sep 04 01:06:02 2001
6315 messages in chronological order