SORT BY:

LIST ORDER
THREAD
AUTHOR
SUBJECT


SEARCH

IPS HOME


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

    Re: iscsi: 08 Comment, framing




    Dave,

    This was the consensus of the framing team - not of the IPS list.
    It is inappropriate for us to replace a thing that we have and is agreed to something that
    that is not here yet.  

    As I stated if you want to do it independent of the ULP progress you have to reopen the issue on this list.

    Regards,
    Julo


    Dave Sheehy <dbs@acropora.rose.agilent.com>
    Sent by: owner-ips@ece.cmu.edu

    09-10-01 08:54
    Please respond to Dave Sheehy

           
            To:        ips@ece.cmu.edu (IETF IP SAN Reflector)
            cc:        
            Subject:        iscsi: 08 Comment, framing

           


    Since the following is the consensus of the framing team:

    > iSCSI, with minor modifications since the presentation at the IETF:
    >
    >
    >                  The Sender:
    >                                   - MUST support no framing
    >                                   - MUST support at least one framing solution
    >                                   - MUST use the framing specified by the receiver,
    >                                                    if it supports that framing mode

    and ...

    >
    > First, a quick summary of the issues:
    >                  - Deploying iSCSI sender can be done:
    >                                   - on top of an unmodified TCP stack with:
    >                                                    o No framing
    >                                                    o Marker based framing
    >                                   - on top of a modified TCP stack can implement
    >                                                    PDU-alignment.
    >                  - The receiver trade-offs are:
    >                                   o No framing - large receive reassembly buffer
    >                                                    (higher cost solution)
    >                                   o Marker framing - receive reassembly buffer is reduced
    >                                                    to an eddy buffer, but requires modifications to
    >
    >                                                    TCP receive stack (medium cost solution)
    >                                   o PDU-alignment - receive reassembly buffer can be
    > reduced
    >                                                    even further, but requires modifications to TCP
    >                                                    receive stack (lowest cost solution and enables
    >                                                    eventual deployment of a viable chunking protocol).
    >                  - The cost of implementing markers on unmodified TCP stacks
    >                                   o sender cost is acceptable, assuming a gather list is
    >                                                    reasonably implemented.
    >                                   o receiver cost is unacceptable

    > Initial implementations for initiator appear to be in two camps:
    >                  - Unmodified TCP software stacks
    >                  - Embedded TCP offload in the NIC (essentially TCP
    >                                   is hidden from the host SCSI stack)
    >
    > Initial implementations for the target appear to be in two camps:
    >                  - Optimized NICs which will support framing - probably both
    >                                   framing modes. Because both modes
    >                                   require modifications to the receive TCP stack and
    >                                   PDU-alignment is viewed as straightforward, it is
    >                                   assumed that most implementations will implement
    >                                   both framing solutions to allow them to transfer
    >                                   data optimally with either unmodified send TCP stacks
    >                                   or PDU-alignment send TCP stacks.
    >                  - Unmodified TCP software stacks
    >
    > It is primarily a receiver cost issue that motivates the framing
    > discussion. The primary sender issue is what, if any, optimizations can
    > be done for unmodified TCP send stacks? Note however, that the proposed
    > iSCSI requirements are not target or initiator oriented - they are
    > sender and receiver oriented. The receiver cost issue applies to either
    > a target or initiator - and they each have very different cost
    > trade-offs.
    >
    > Thus the best solution is to allow the receive side to control their
    > destiny - and require the sender to obey the receiver (within limits).
    > If this approach were taken, the sender would have to implement all
    > three framing options. Unfortunately, a highly desirable design goal is
    > to allow the sender (either the target or the initiator) to run on an
    > unmodified TCP stack. Thus the compromise in terms of sender
    > functionality.

    The iSCSI 08 draft Appendix C MUST be amended,

    from,

    "The use of markers is negotiable. The initiator and target MAY
    indicate their readiness to receive and/or send markers during login
    separately for each connection. The default is NO. In certain
    environments a sender not willing to supply markers to a receiver
    willing to accept markers MAY suffer from a considerable performance

    degradation."

    to:

    "The use of markers is negotiable. The initiator and target MUST
    indicate their readiness to receive and/or send markers during login
    separately for each connection. The default is NO. In certain
    environments a sender not willing to supply markers to a receiver
    willing to accept markers MAY suffer from a considerable performance
    degradation."

    and in the next paragraph:

    "If a receiver indicates that it desires a marker, the
    sender SHOULD agree (during negotiation) and provide the marker at
    the desired interval."

    MUST be changed to:

    "If a receiver indicates that it desires a marker, the
    sender MUST agree (during negotiation) and provide the marker at
    the desired interval."

    Dave Sheehy





Home

Last updated: Tue Oct 09 21:17:31 2001
7169 messages in chronological order