FCIP: Jna 2 2002 Teleconf Call Minutes

["Murali Rajagopal" <muralir@lightsand.com>]

Minutes: Taken by Don Fraser, Compaq

FCIP Con-Call Wednesday, January 02, 2002 1 PM PST

Attendance:

Murali Rajagopal,
Bill Krieg,
Andy Helland,
Anil Rijhsinghani,
Jim Nelson,
Venkat Rangan,
Milan Merhar,
Raj Bhagwat,
Ken Hirata,
David Peterson,
Bob Snively,
Larry Lamers,
Ralph Weber,
Don Fraser

Agenda with notes:

1. Short Frame Proposal Comments: 30 Min

Discussion on email comments from Dave Peterson and Bob Snively re the
various SF proposals.  The discussions confirmed:

a) FCIP document needs to specify requirement for a random nonce, but not
the method.  The implementation should not allow it to be possible to guess
the next number, see RFC 1750.  Note RFC is a 30 page  informational RFC
that on first glance looks harmless.

b) Will expand the FC entity field size to 64 bits.  This new space will
hold a port name but should not be required to be the port name.  There
should be no checking of the values because the B-port model does not have a
port name.   Carry forward to BB-2 as well.

c) There is no reason to specify how to name the FC Entity.

Discussion around adding one or more reserved words back into the SF for
example to allow for a CRC.  Impact may be that one implementation uses it
and is therefore non-interoperable with those that do use them.  Consensus
was to not add in any reserved words with Larry objecting.

Version number of protocol to be version number of short frame.

There was agreement that the first connection must be authenticated prior to
starting any additional connections between the pair of devices.

To prevent a possible race condition in the creation of a connection, the
document will require that duplicate nonces, and re-use of nonces on new
connections must be prevented.  Ralph will write it up.

Will change name from Short Frame (SF) to Special Frame (SF)


2. Comments on 07 draft: 20 Min (moved from item 3 to 2)

Don's comments by page:
In general, drop all added timers, etc as they belong in FC-BB-2
page 20, #1 accepted, #2 drop timer, #3 drop
page 25, #1 drop timer, #2 ok
page 30, shall provide mechanism with acknowledgement
	page 31, #1 move to FC-BB2, add reference to FC-BB-2, #2 ok
	page 37, both add only the notification with reason
	page 48, add only the notification with reason code
	page 50, add the reason code and notifications

para 8.2 discussion and will be further discussed at a later date due to
implication for path split, some would delete the paragraph others would
require it.
*	Agreed that there is a one to one correspondence between the link end
point, its data engine, and the TCP connection direction.
*	May need to delete it, as there is no time to define how to combine two
links, even if it could be proved that this is needed.  Bob to forward some
rational for dropping 8.2.
*	Need to have FC-BB tell the FCIP device to build a new connection rather
than figure out how to tell the FCIP side to merge two connections.
*	Consensus to remove, 8-2

Venkat brought up that there are some potential timing issues with section 9
due to delays in the common security specification from the IPS Security
Working Group.  It was decided to proceed and then if the IPS side is not
ready, so be it.

Class 4 list to be sent to Gary Stevens and ask him to review.

3. Discussion on FCIP Requirements to FC-BB-2: 30 Min

	- deferred to next week due to length of discussion on item 2.

4. Gadzoox to plan Next Call


Don Fraser
Solutions Architect
Wide Area SAN Engineering Team
Compaq Enterprise Storage Products Group
Phone: 719.548.3272
Pager: 1.800.759.8888 pin 114.8840 or
      short email to 1148840@skytel.com