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

Re: AW: Deconstructing OAM&P




Vipul,

The single biggest difference between native LAN protocols such as 802.3
and native WAN protocols such as SONET/SDH or local loop D1/E2/D3, is
that the WAN protocols provide an operations service functionality.  For
example, on a T1/D1 service between a user's DSU/CSU and the service
provider multiplexer, the operations overhead gives the service provider
the ability to check for excessive bit errors, bad wiring, excessive
latency in one direction, data jitter and timing errors, encoding
errors, etc. without impacting the user's data.  Often these operations
support functions are done without the user being aware of them.

In the case of optical transport systems, the same service support
functionality is used.  Optical transport systems such as SONET/SDH have
different levels of support functionality based on path, line, and
section.  There are individuals that work for transmission systems
vendors that can give you more detail on the specifics of SONET and
SDH.  The path level of support functionality can be considered as
between one user end system and another user end system.  For 10GbE only
the path overhead processing should be used.  The single biggest benefit
is that it gives a BER directly related to the optical signaling,
allowing the detection of fiber transmission quality issues directly.
It also provides a interface port to interface port trace function,
providing the ability to identify the link without looking at the data.
There are other path level support functions as well that are good to
have for WAN systems.  Line and section overhead processing, can be left
to the carrier transmission or DWDM systems.  In a concatenated
SONET/SDH system, the bulk of the OAM&P processing is at the line and
section level, not needed for 10GbE.

10GbE would look like a concatenated OC192/STM64 to the SONET/SDH
transmission systems.  The path overhead for concatenated SONET and
concatenated SDH are the same, making 10GbE universal over the
transmission systems of the U.S.A., Canada, Europe, and other parts of
the world.  This is not true of WAN services today.  This gives a native
10GbE, that leverages the SONET/SDH technology, the ability to become
the universal data protocol, end to end, all over the world.  Only with
a native 10GbE protocol, that leverages the SONET/SDH technology,  802.3
frames can be delivered from anywhere in the world to anywhere in the
world without modification.  802.3 would truly become the universal
protocol.

Thank you,
Roy Bynum
MCI WorldCom



Vipul Bhatt wrote:

> Hello,
>
> While I consider myself a member of the LAN Camp, whatever that
> means, I also think we will suffer some kind of a loss by completely
> rejecting Roy Bynum's arguments.
>
> If I understand Roy correctly, he is not arguing in favor of making
> Ethernet embrace WAN "with all of its baggage"; on the contrary, he
> has suggested a pruned down version of it.
>
> Every camp has a set of tenets. Many of us easily state that the LAN
> camp's tenet is "keep it simple, keep it cost-effective". I believe
> one of the telco camp's tenets is "keep it manageable." Since I
> strongly believe in my tenet, I must concede that there must be
> reasons why the other person believes in his tenets.
>
> Roy, may I make a suggestion. If you believe that Ethernet as is
> (i.e., without any OAM&P functions) will not succeed as a WAN
> solution, please educate us as to why not. Please take specific
> examples - a cable broke, a connector got dirty, power to a switch
> hut was lost.....I am so naive I can't even think of examples - and
> educate us as to why the additional cost of some of the Path
> management functions you have suggested is not as high as the
> benefits are.
>
> Thank you.
>
> Vipul
>
> Vipul Bhatt
> Finisar Corporation
> 274 Ferguson Drive
> Mountain View CA 94043
> Phone:(650)691-4000 x113
> Fax:(360)397-8963
> Email: vipul.bhatt@xxxxxxxxxxx
>
> >
> > Quite frankly, I find the argument made by a few people
> > that the mere fact that
> > we are using fiber optics as a transmission medium for
> > Ethernet somehow leaves
> > us with no other choice but to embrace the WAN with all
> > of its baggage, utterly
> > ridiculous. Following this philosophy, if we decide to
> > run Ethernet over barbed
> > wire, we will have no choice but to move 802.3 under the
> > auspices of the
> > Department of Corrections...
> >
> >
> >                                                       Shimon.
> >