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

RE: XGMII a/k/r




Ben,

Yes, /lowercase/ meant 8b and /uppercase/ meant 10b, except for on the
medium.  Like I said in my note, I left them as uppercase because I wasn't
about to invent a new nomenclature for the medium.  But, to avoid confusion,
it might be a good idea to address that.

The /o/ was left in for the purpose of OAM&P as Osamu Ishida proposed in
XGENIE.  At first, I wasn't into using the /o/d/d/d/ in the idle stream.  As
I thought about it some more, I realized that if the LAN PMDs could transmit
up 80 or 100 km, then there was the possibility that the source and
destination were more than a few 100 kms apart.  If that was the case, what
guarantee was there that this was not over leased dark fiber. If the
regenerators on that leased line were not SONET regenerators, but
regenerators based upon the LAN PHY, then having a Layer 1 OAM&P could be
extremely useful.  Remote fault is part of OAM&P.  Break link is an
auto-negotiation thing and is not required if we don't have
auto-negotiation.

To define all the OAM&P capabilities would not be an easy add to the
standard, and maybe should be left to another standard development or
consortium effort.  If we could write the standard to filter it, then future
efforts could work to define it.

Of course, this is a moot point if all the companies in the world feel that
they'll only use SONET from long haul communication over lease fiber (dark
or lit), because in that situation, SONET already supplies OAM&P.

Cheers,
Brad

		-----Original Message-----
		From:	Brown, Ben [BAY:NHBED:DS48]
[mailto:bebrown@xxxxxxxxxxxxxxxxxx]
		Sent:	Thursday, April 06, 2000 6:58 PM
		To:	HSSG
		Subject:	Re: XGMII a/k/r



		Brad,

		Okay. If I take your description literally, the /lowercase/
means 8b
		octet while the /uppercase/ means 10b code-group. Does this
mean the
		10b code group is transferred across the medium? I don't
think this
		has ever been suggested, at least since we discounted 12.5
Gbaud
		optics.

		Aside from above, I have a question about the /o/ (I use
lower
		case because these are carried from PCS to PCS (PHY to PHY?)
in
		your picture). Would (Could?) these be used to carry your
/rf/ &
		/bl/? Other than the OAM&P that Osamu Ishida has suggested
(XGENIE)
		and Fibre Channel "things" (both of which deserve more
discussion),
		what else would the /o/ carry? Why would /rf/ & /bl/ use
dedicated
		encodings? This might allow for multiple reasons for /rf/
and /bl/
		or even count values or something else clever within the
/o/d/d/d/
		ordered set.

		Ben