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

Re: [EFM] RE: [EFM-P2MP] 10G EPONs (it was MPCP: Report message)


In the asymmetric versus symmetric bandwidth for Internet access and any 
other type of interactive communications, I would come down on the 
symmetric side of the argument.  I would however point out that Internet 
access is often as much as a 400 to 1 statistical gain environment.  It has 
to be that way in order for the bandwidth to be solvent at the commodity 
prices that it is going for.  That kind of statistical gain 400/1 does not 
work well in a symmetrical service, particularly during busy hour 
usage.  The gain factor for interactive "Internet" type bandwidth services 
has to change, but that means the overall cost to the customer has to 
change as well.

There is a way out of this impossible paradox of needing to provide a 
service in order to gain customers, but unable to make the service pay for 
itself.  That is to be able to provide other services, such as high 
bandwidth video content, to the same customer over a common 
infrastructure.  This allows the infrastructure costs for the service 
facilities to be shared by multiple services.  It also improves the "take 
rate" for the one or more of the services over that common infrastructure, 
which has an additional economic benefit.  This requires an infrastructure 
that can support unidirectional service facilities as well as symmetrical 
bidirectional service facilities.

Thank you,
Roy Bynum

At 11:24 AM 2/13/2002 -0200, Carlos Ribeiro wrote:

> > Assuming we agree with you with the amount of bandwidth that is needed
> > in the downstream. For this 2.5 Gbps are good enough. In that case why
> > we need to pay for expansive optic in the upstream. Home users will
> > still need the same small amount of bandwidth. 622 will be enough. I
> > guess asymmetric rates are making a lot of sense for FTTH application.
> > B.t.w - The FSAN ILEC's are very consistent with their asymmetric rates
> > requirements.
>Assymetric rates are an artifact from the common perception that users want
>high speed content fed upon them. It's also related to the fear that MSOs
>will dominate the access network, so ILECs need a way to counteract and to
>provide CATV-equivalent services. However, this is not a consensus, even
>inside the biggest ILECS.
>There has been considerable debate whether assymetric rates are the best way
>to deploy a broadband network. I have actual experience on this, because I'm
>responsible for the supporting IP backbone at CTBC Telecom; there we are
>seeing a growing trend towards symetric usage, even from otherwise normal
>websurfers. It was very common to have ratios such as 4:1
>upstream/downstream; now, it's getting close to 2.5:1. What is more
>interesting - the difference is smaller with DSL and cable modem users,
>around 2:1 (or even less!).
>Some lessons we've learnt:
>- Content is not king. People want to communicate. WAP was a flop, but SMS is
>a winner. The screen size and the keyboard are the same, but one is 'central
>content', the other one is 'one-to-one communication'.
>- People like to share content. Legal problems aside, that's the fact.
>- Hosting content is not hard. In fact, it will only become increasingly
>Another reason *not* to use lower speed optics on the upstream for EFM P2MP:
>even if the optics are the same, the system is already assymetric. Using
>lower speed optics at the upstream will only make things worse.
>Carlos Ribeiro
>CTBC Telecom
>Carlos Ribeiro
>CTBC Telecom