Re: [8023-10GEPON] Asymmetric MII thoughts
Duane,
Thank you for volunteering -- asymmetric MII is an important focus area with
very little work done so far.
Few comments:
1) If the group decides to go with approach #1 (defining two new
bi-directional interfaces), it may be better to use the same designation D
and U as we used for PMDs in 1G EPON.
2) Under "Something else" category:
What if we define RS and PCS sublayers that have both GMII and XGMII
interfaces between them? In the OLT, RS with forward outgoing data to XGMII
Tx path, but will read incoming data only from GMII Rx path.
The RS and PCS sublayers at the ONU will do the opposite: RS with forward
outgoing data to GMII Tx path, but will read incoming data only from XGMII
Rx path.
This way, we don't need to touch GMII and XGMII or define any new MIIs. We
can only define a new RS and PCS sublayers, which based on a configuration
parameters would act as either 10G-Rx/1G-Tx or 1G-Rx/10G-Tx.
Glen
> -----Original Message-----
> From: Duane Remein [mailto:duane.remein@ALCATEL.COM]
> Sent: Friday, November 17, 2006 10:33 AM
> To: STDS-802-3-10GEPON@LISTSERV.IEEE.ORG
> Subject: [8023-10GEPON] Asymmetric MII thoughts
>
> Gents,
> After Eric's excellent presentation on Asymmetric Media Independent
> Interface I realized that we have at least two interfaces, one for the
> OLT and another for the ONU. As I see it we could do this a number of ways
> 1) Define two bi-directional interfaces;
> one Asymmetric MII 1Gb/s ingress 10Gb/s egress for OLT (AMII-L)
> the other 10Gb/s ingress 1Gb/s egress for ONU (AMII-U)
> 2) Define four uni-directional interfaces;
> GMII Rx (GMII-R)
> GMII Tx (GMII-T)
> 10 Gigabit Attachment Unit Interface Rx (XAUI R)
> 10 Gigabit Attachment Unit Interface Tx (XAUI T)
> 3) Something else?
> We also need to agree on naming for these interfaces (my suggestions are
> in paren's above). I'm not wild about AMII-L and AMII/U but it's better
> than AMII-101 and AMII-110. We had similar discussion about naming in
> EFM days about the PON Phys, you may want review that work.
> I will volunteer to work up a follow-on presentation on how to
> slice/dice and reference existing specs as we discussed in Dallas, if we
> can come to a timely agreement on how we want to approach this.
> Any thoughts?
> Duane