[802.3_DMLT] Minutes for IET baseline ad-hoc

["Hugh Barrass (hbarrass)" <hbarrass@xxxxxxxxx>]

All,

 

Belated minutes for the baseline ad-hoc meeting that was held last week.

 

Attendees: Hugh Barrass, Pat Thaler, Arvind Kansal, Brian Allen, David Chen, Albert Tretter, Denis Beaudoin, Anatoly Moldovansky, Peter Jones

 

Hugh reminded attendees of the patent policy and displayed the links to access the policy online.

 

The discussion continued to focus on the baseline presentation adopted by the Task Force:

 

http://www.ieee802.org/3/br/8023-IET-TF-1401_thaler-01-0114-iet-proposal-v2.pdf

 

Pat summarized the discussion that happened off-line with Mick Seamen and some background view of MACsec. Out-of-order is allowed, with a pre-determined limit to the number of packets re-ordered. It should be expected that most implementations would use a single copy of MACsec (with state saved for a preempted packet). This was considered preferable to having 2 instances (with 2 key negotiations, etc.).

 

CRC inversion - it is preferred no to invert the whole CRC as that would undo the inversion that is defined in 802.3 to prevent a corner case where a frame ends in zeroes & runs into a CRC of all zeroes (unclear whether this is a real problem).

 

There may be some opportunities to make the SFD changes friendlier to (widely used) implementations that use some non-standard preamble behavior. These should be investigated & discussed further. The next call is scheduled for 8.00am PDT on April 24th, it would be useful to discuss a complete proposal that covers an alternate scheme for start frame delimiting so that a fair comparison can be made.

 

Hugh.