RE: BOF Request

[julian_satran@il.ibm.com]

Mark,

I am part of another organization that has some work going on in both the
storage and networking areas (as are you).  I think that Costa is right in
assuming that most of us know the area and the requirements could be defined
very fast.  It would be a pity not to move to something more substantive during
this meeting - as the area is mature and badly in need of good solutions.

Regards,
Julo

Julian Satran - Senior Technical Staff Member - IBM Research

"Bradley, Mark" <mark_bradley@btc.adaptec.com> on 11/02/2000 23:28:23

Please respond to "Bradley, Mark" <mark_bradley@btc.adaptec.com>

To:   "'Costa Sapuntzakis'" <csapuntz@cisco.com>, ips@ece.cmu.edu,
      dave_lee@3com.com
cc:    (bcc: Julian Satran/Haifa/IBM)
Subject:  RE: BOF Request




This presumes that everyone knows and understands the issues
and problems from the various perspectives that may be provided.
Since this is not the case, we need to provide a venue to those
to be communicated and discussed.  Once we have some common
understandings of the issues, perhaps we can work on requirements.

Then, and only then, would it make sense to entertain proposals
that satisfy the agreed upon requirements.  I believe we should
attempt to do as thorough a job as possible in understanding the
issues and defining requirements as is humanly possible.  Only
then can we evaluate proposals against needs.

This might fly in the face of TTM (time to market) for someone's
product, but it is the best method I can imagine to develop
standards that meet the needs defined by the industry.

Let's follow the process that the IETF has laid out for this work.
  --  markb

> -----Original Message-----
> From: Costa Sapuntzakis [mailto:csapuntz@cisco.com]
> Sent: Thursday, February 10, 2000 3:55 AM
> To: ips@ece.cmu.edu; dave_lee@3com.com
> Cc: djpotter@cisco.com; cheriton@cisco.com; avb@cisco.com;
> vern@aciri.org; sob@harvard.edu
> Subject: Re: BOF Request
>
>
>
> Hi,
>
> I'm working with Andy Bechtolsheim on the IP SAN initiative at Cisco.
> I believe we (the industry) are behind at this point because we
> (Cisco) have been hearing from our customers (and I'm sure you've been
> hearing from yours) that they want these technologies now. As such,
> much of the discussion on the charter and goals should happen before
> the BOF (on the ips mailing list).
>
> Let's make the IETF meeting be for serious high-bandwidth engineering
> discussion of the various approaches so we can attempt to reach
> consensus on a technical direction.
>
> Proposed agenda for BOF:
>
> > 5 min Review of BOF agenda and objectives (D. Lee/M. Bradley)
> > 10 min IP Storage Goals (D. Nagle/A. Wilson)
> > 15 min Discussion
> > 30 min Approaches & Implementations (10 min presentations)
> > o SCSI/TCP (TBA)
> > o Adaptec (A. Wilson)
> > o ISI's NetSCSI (?)
> > 40 min discussion
> > 5 min Proposed charters (D. Lee/M. Bradley/C. Sapuntzakis)
> > 15 min Discussion
>
> Our formal interaction with T10 and T11 should be minimized. The IETF
> strongly discourages the republishing of IETF standards by other
> organizations. In addition, IETF can work much more quickly than
> ANSI standards organizations.
>
> The only interaction I see with T13 is if we define an IDE
> transport protocol.
> ATAPI command sets are now handled through T10.
>
> FCP is an impressive protocol, especially in the bulk of documentation
> required to understand it. It is very much tied to Fibre Channel
> Physical Layer. Understanding FCP, its limitations, how it and its
> associated discovery mechanisms need to be fixed up to work in an IP
> environment is a large undertaking. FCP will seriously slow down the
> generation of any standard and impede the clarity of the final
> standard.
>
>
> A proposed charter:
> -------------------
>
> Description of Working Group:
>
> The goal of this working group is to define extensions to current IETF
> transports for storage networking.
>
> There are a several areas to consider:
>
>     - Reducing protocol processing overhead at servers and clients
>     - SCSI transport on top of TCP
>     - QoS parameters for storage networks
>     - Running storage transports at multi-gigabit speeds
>
> The objectives of this working group are:
>
> * A proposed standard for a SCSI transport on top of TCP
>
> * A proposed standard for extensions to TCP that reduce protocol
> processing overhead.
>
> * Produce an informational RFC describing the QoS
> requirements of storage
> networks. Propose QoS extensions to diffserv if necessary.
>
> * Produce an informational RFC discussing the issues behind running
> storage transports at multi-gigabit speeds.
>
> --------------------------------------
>
> -Costa
>
> >
> > ----- Begin Included Message -----
> >
> > >From Dave_Lee@3com.com Wed Feb  9 03:17:02 2000
> > From: Dave_Lee@3com.com
> > X-Lotus-FromDomain: 3COM
> > To: agenda@ietf.org
> > cc: sob@harvard.edu, mankin@isi.edu, Vern@Aciri.Org,
> markb@btc.adaptec.com,
> >         Carl_Madison@3com.com, avb@cisco.com
> > Date: Wed, 9 Feb 2000 03:15:23 -0800
> > Subject: BOF Request
> > Mime-Version: 1.0
> > Content-Disposition: inline
> > X-SMTP-HELO: seattle.3com.com
> > X-SMTP-MAIL-FROM: Dave_Lee@3com.com
> > X-SMAP-Received-From: outside
> > X-SMTP-PEER-INFO: seattle.3com.com [129.213.128.97]
> >
> >
> >
> > BOF Description
> >
> > The objective of this BOF is:
> >
> > 1. To consider the creation of a new IETF working group to
> enable storage
> > networking via IP-based storage interfaces of SCSI, ATAPI,
> FCP and other
> > interfaces.  Only system-to-storage interfaces will be considered.
> >
> > 2. To develop and agree upon a charter for the new working group.
> >
> > 3. To consider the proposed operational characteristics of
> reliable and
> > unreliable communications, shared and non-shared, managed
> and unmanaged, and
> > integrity, privacy, and/or non-secure.  Further, the protocol must
> > efficiently operate over LAN, campus, and WAN environments.
> >
> > 4. To consider proposed protocol features such as automatic
> discovery,
> > enumeration, configuration, and management of storage
> adapters and devices.
> >
> > 5. To discuss what collaboration with ANSI T10, T11 and T13
> is appropriate.
> >
> >
> > BOF Agenda (2 hrs)
> > 5 min Review of BOF agenda and objectives (D. Lee/M. Bradley)
> > 20 min IP Storage Goals (D. Nagle/A. Wilson)
> > 35 min Discussion
> > 20 min Approaches (5 minutes each)
> > o Cisco (A. Bechtolsheim)
> > o Adaptec (A. Wilson)
> > o ISI's NetSCSI (?)
> > o Other (?)
> > 10 min Proposed charter (D. Lee/M. Bradley)
> > 30 min Discussion
> >
> >
> >
> >
> >
> >
> > ----- End Included Message -----
> >
> >
> >
>