SORT BY:

LIST ORDER
THREAD
AUTHOR
SUBJECT


SEARCH

IPS HOME


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

    RE: not every song



    
    
    I any case, while we are here to discuss iSCSI, we have the both.  Julo
    
    Robert Reynolds <robertr@Crossroads.com> on 30/09/2000 00:31:16
    
    Please respond to Robert Reynolds <robertr@Crossroads.com>
    
    To:   "'ips@ece.cmu.edu'" <ips@ece.cmu.edu>
    cc:    (bcc: Julian Satran/Haifa/IBM)
    Subject:  RE: not every song
    
    
    
    
    
     I agree with Douglas.  The WRITE-DATA and DATA-RESPONSE
     message become more important when you are talking longer
     distances, and longer delays.  You can reduce your roundtrip
     latency by 1/2 on WRITE commands by using a WRITE-DATA message
     instead of independent WRITE-XFER_RDY-DATA messages.  This
     can be a huge performance increase when talking about doing
     data replication across longer latency connections.  This is
     the difference between a 2 hour database sync. delay and a 4
     hour sync delay.
    
     I think it is a mistake that FCP-2 removed those messages because
     they weren't currently implemented and I think it would be
     a mistake to remove them from iSCSI.
    
       Bob
    
    > -----Original Message-----
    > From: Douglas Otis [mailto:dotis@sanlight.net]
    > Sent: Thursday, September 28, 2000 3:50 PM
    > To: Robert Snively; John Hufferd/San Jose/IBM; ips@ece.cmu.edu
    > Subject: RE: not every song
    >
    >
    > Bob,
    >
    > I am not sure I understand the point.  Not that I would advocate such
    > modifications, but should these points become a deciding
    > factor, the service
    > parameter for such negotiations has already been defined
    > within FCP-2 as of
    > recently.  Such a decision for removal may have been premature should
    > network latency argue for Word 3, bit 3 being set.
    >
    > Again, I am agnostic about this issue, but I would rather
    > stick with FCP
    > structures.  Should there be attempts to provide zero copy,
    > adopting FCP
    > structures permit a greater reliance on these structure's stability.
    >
    > Doug
    >
    > > -----Original Message-----
    > > From: Robert Snively [mailto:rsnively@Brocade.COM]
    > > Sent: Thursday, September 28, 2000 12:37 PM
    > > To: 'Douglas Otis'; John Hufferd/San Jose/IBM; ips@ece.cmu.edu
    > > Subject: RE: not every song
    > >
    > >
    > > >  FCP
    > > >  offers WRITE-DATA and READ-RESPONSE structures if desired.
    > >
    > > Reminder:
    > >
    > > No, it doesn't.  FCP-2 (the new and improved FCP) took the
    > > function out because nobody implemented it and it did not make
    > > sense.  It would have required additional recovery mechanisms and
    > > it did not improve performance.
    > >
    > > Bob
    > >
    >
    
    
    
    


Home

Last updated: Tue Sep 04 01:06:56 2001
6315 messages in chronological order