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

RE: [802.3ae_Serial] - link model modification requests




Piers,
on the call today you mentioned that you were considering adding an input
cell to the link model to account for splitter loss in EFM. You suggested
that this cell could also serve the purpose of defining the additional loss
allowed for 10G PMDs and thereby indirectly resolve my request for a new
cell to allocate reserved margin. 

I've had a chance to think about the solution you suggested and believe it
would be a sub-optimal compromise. While I agree that such a cell could
serve both purposes, it has a drawback in the case of substituting for a
reserved margin cell. The drawback stems from the fact that we have chosen
to apply a fixed amount of reserved margin to each PMD type across all fiber
types that support it. The  resultant additional insertion loss allowance
varies by fiber type. Thus, while a single value applies for reserved margin
for each PMD across all fiber types, with the approach you suggested
multiple values are needed for additional loss allocation, one for each
fiber type. This could require a separate worksheet for each PMD/fiber
combination, a complexity that I would rather avoid. It also means needing
to tweak the additional loss entry to a value that results in the same
margin at the specified link length for all fiber types supporting a
particular PMD. I'd rather have the link model do the calculations. It's
less obscure and less error prone. 

I think a cleaner approach would be to insert two new cells, one for each
purpose. The additional loss cell would be used to allocate a fixed amount
of loss (e.g. for splitters). The reserved margin cell would be used for
making a portion of the power budget unavailable. The present "margin"
column could be re-titled to "available margin", which under present
philosophy would be used to fill in 10GbE table entries described as
"additional insertion loss allowed". 

I hope this explanation makes my points sufficiently clear. If not, or if
you have another view, please respond.

Regards,
Paul Kolesar