Ronen,

The problem is that in 1000BaseT you need to use the same clock in both 
directions
(earlier copper physical interfaces used a different clock in each directions,
but with Gb that would lead to cross talk degrading the performance).

802.3-2005 states that when you use autonegotiation, the master is chosen 
randomly.
In fact, a random or pseudorandom number is generated for this purpose.

This would seem to prohibit the use of Gb copper for SyncE,
as the direction of the clock flow has a 50% chance of ending up wrong
(i.e. "upstream").

However, apparently the 802.3 people in their wisdom foresaw this problem.
They defined a control register bit (9.12) which forces a device to be master.
If both sides set this bit the autonegotiation fails, so if the link comes up
you can be sure the right side is master.

Y(J)S



From: [email protected] [mailto:[email protected]] On Behalf Of 
Ronen Solomon
Sent: Tuesday, March 24, 2009 04:33
To: [email protected]
Cc: Rafi Ram
Subject: [TICTOC] bi-directional SyncE - Copper GBE

Yaakov

In the last meeting you raised the issue of supporting bi-directional SyncE on 
Copper GBE. Can you please re-emphasis the problem and the proposed solution.

Best regards

Ronen
_______________________________________________
TICTOC mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/tictoc

Reply via email to