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

Re: [10GBASE-T] Objectives


At 03:22 PM 7/30/2003 -0600, wrote:
>I think one area where the objectives need work is with regard to MAC 
>compatability. The only one I see that currently relates to it is:
>Support a speed of 10.000 Gb/s at the MAC/PLS service interface
>Given some of the proposals being made, I don't think this objective goes 
>far enough. (I note that this doesn't even say that 10 Gb/s is the only 
>speed that will be supported.)  In the past we have occasionally tweaked 
>the MAC definition to accomodate faster operation. When we are adding a 
>new physical layer for an existing MAC speed, we have always been careful 
>to design the new PHY so that it was entirely compatible with existing 
>MACs. For instance, when we designed 10BASE-T, we made sure that the 
>behavior the 10BASE-T MAU exhibited when a fault occurred matched behavior 
>that coax MAUs produced for faults.
>I propose adding an objective:
>Compatible with existing MACs at XGMII and XAUI interfaces without 
>If there are people who don't think that should be an objective,

I believe that 802.3 should be using this sort of criteria as part of its 
internal review of the "Compatibility"

That is, assuming that it would be compatible at 802.3's usual points would 
be a baseline assumption and those who propose new projects would have to 
declare their intention/ability to be compatible at the points we care 
about as opposed to the points that are specified in the 5 Critters. Any 
departures would have to be explicitly called out and agreed to by 802.3

>then they should explain what sort of modifications they expect to need 
>and how a PHY that is not compatible with existing MACs (including switch 
>ports) can have broad market potential.

In other words, Pat and I are in violent agreement here.