Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: Break Link and Remote Fault

Shawn, Dineen,

At 12:29 00/07/17 -0500, Rogers, Shawn wrote:
> I agree there was some FUD implying LSS was a form of AN, 
> however the main point I heard against LSS was it's use of 
> Inter Packet Gap (IPG).  The 802.3 has resisted the use of 
> IPG for anything, time and time again.  I do not see positive 
> movement on supporting LSS as long as it is done in the IPG.  

At 11:06 00/07/17 -0700, Thomas Dineen wrote:
> 	Did it fail because of opposition to the concept of
> signaling Break Link and Remote Fault, or did it fail because
> of opposition to the proposed method of signaling in the IFG?

Thank you for illuminating the main opposition to the LSS 
in the La Jolla meeting; the use of IPG. 

To tell the truth, as a full-duplex Layer-1 (SONET) guy, 
I have not yet fully understood what issues are there.  
So it would be appreciated if you could help me to figure out 
what is the real argument to the current LSS proposal.

The LSS is the PHY signaling mechanism using the IDLE column 
regardless of the presence or absence of the Ether Frame.
As a presenter of the above proposal, I admit my misleading 
words that may highlight the IPG usage too much.  It uses the 
IDLE column, not the IPG itself,  for Layer-1 signaling.

At present I myself don't see any reason for 10GbE PHY to 
distinguish the IDLE columns in IPG from those in pure IDLE 
stream with no Ether Frame.

As I understand, Ethernet Break Link and Remote Fault is a 
required Layer-1 signaling whose direct clients are PHY itself or 
MDIO, not the MAC nor XGMII.  Therefore, it is natural to me 
that PHY does not care about the XGMII status; the presence 
or absence of the Ether Frame.  For BL and RF signaling, the PHY 
is simply waiting the IDLE column to be replaced.

Restricting the Layer-1 signaling during the pure IDLE stream 
alone reminds me the Auto-Negotiation process; special link 
start-up sequence before starting the Layer-2 communication.

I guess this Auto-Negotiation or assuming the L1 signaling 
during the link startup alone is due to CSMA/CD or half-duplex 
nature of the traditional Ethernet.  However, 802.3ae has 
decided to provide full-duplex link alone, and that's why we 
will have no Auto-Negotiation process.  In this situation, 
I believe it is more simple that the PHY does not care about the 
XGMII status when the IDLE column carries the L1 signaling 
information, such as RF and BL.  

In short, in full-duplex point to point link, I see no reason 
for PHY to distinguish the IDLE columns in IPG from those in 
pure IDLE streams.

I might be missing something; so please let me know if you can 
find some inconsistency in my current understanding.

Best Regards,

NTT Network Innovation Laboratories
TEL +81-468-59-3263  FAX +81-468-55-1282