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

[EFM-P2MP] Upstream efficiency



Title: RE: [EFM] [EFM-P2MP] 10G EPONs
David,
 
Ethernets do not suffer a 20% loss in efficiency due to 8/10 coding as implementation loss is counted from the 1Gbps bit rate, not the 1.25Gbps line rate.
I also think your calculation for burst overhead is severe, and not in line with the 1micro-second guard band that is being promoted.
Overall I would say that upstream efficiency can easily be higher than 80% (and not the 50% presented).
 
I would also like to isolate upstream efficiency induced by link limitations, such as laser performance, from MPCP efficiency.
I believe there is no penalty associated with the proposed protocol, that is in line with Ethernet in general. Upstream utilization will be maximized based on the performance of the optics.
 
Best regards,
    Ariel
-----Original Message-----
From: David Levi [mailto:owner-stds-802-3-efm-p2mp@majordomo.ieee.org]On Behalf Of David Levi
Sent: Wednesday, February 13, 2002 11:58 PM
To: carlosal@xxxxxxxxxxxxxxxxxx; gerry.pesavento@xxxxxxxxxxxx
Cc: bob.barrett@fiberintheloop.com; stds-802-3-efm@ieee.org; stds-802-3-efm-p2mp@ieee.org; vincent.bemmel@alloptic.com
Subject: RE: [EFM] [EFM-P2MP] 10G EPONs
.
  • The MPCP protocol efficiency for 1.25Gbps upstream is 50% (in 1ms cycle). 8/10 coding takes 20%, no fragmentation takes 15 % more, and the burst OH takes additional 15%.  

We did not decide that fragmentation is wrong, it is just because the 802.3 is the host of the EPON standardization activity and 802.3 does not allow to fragment packets. In addition the 802.3 requires 8/10 coding, which result in 80% efficacy.

Regards

David

www.broadlight.com

winmail.dat