RE: FCoverSCTP/IP specification

["Douglas Otis" <dotis@sanlight.net>]

David,

Any FC traffic could be encapsulated.  You may wish to make a separate
informative RFC as to how to use such an encapsulation to exchange IP over
FC, FCP-x, etc.  This could eventually include buffering and bandwidth
management tricks, but with this separate RFC.  Again, I would expect that
to be other documents and likely ones created by T11 themselves.
http://www.t11.org

I think the major trick is to avoid crossing into T11 areas as their work is
ongoing and would be a duplication of efforts to attempt to define some kind
of subset.  This spec should relate to the mundane interconnects,
cross-links and the like.  (Networking.)

Doug



> -----Original Message-----
> From: owner-ips@ece.cmu.edu [mailto:owner-ips@ece.cmu.edu]On Behalf Of
> David Robinson
> Sent: Monday, August 21, 2000 5:05 PM
> To: ips@ece.cmu.edu
> Subject: RE: FCoverSCTP/IP specification
>
>
> In reading the discussion of FC over SCTP there has been a lot of
> high level posturing but a distinct lack of details.
>
> Assuming we all have and understand both the FC specs and SCTP specs,
> there is a whole lot of details beyond just stating "run FC over SCTP".
>
> There are lots of interesting questions that need to be addressed, simple
> examples of which might be:
>
> 1) What level of FC is running over SCTP? FC-0 or FC-4 or something
>    in between? Or just FCP? Or FCP-2?
>
> 2) Of the various FC service classes, which are supported and
> which are not?
>
> 3) Of the various Mode pages defined in SCSI and FCP, how are the values
>    interpreted and what should the recommended values be?
>
> 4) If third part transfers are supported what is the addressing used?
>
> 5) How is login and authentication done?
>
> etc etc etc.
>
> [I really don't want to hear answers to these particular questions
> in response to this e-mail, they are but 5 of many dozens I can think of.]
>
> What I would like to see is the proponents of FCoverSCTP/IP take the
> time to write up a detailed draft that has enough information that
> someone could make an attempt to prototype an implementation.  Without
> such a draft there is no purpose debating the theoretical aspects of an
> undefined protocol. The other IPS protocols, SEP, iSCSI and the FC
> tunneling encapsulation, have such detailed drafts that I can have a
> concrete discussion about. I believe FCoverSCTP/IP may have merit, but
> I can't know until I see details.
>
> So until I see a real draft with real details I can discuss, I am going
> to ignore any further discussion and strongly encourage others to
> do the same.
>
> 	-David
>
>