The MPLS WG Archive

[Curtis Villamizar <curtis@fictitious.org>]

In message <3E64FAE6.AC5BC96@lucent.com>, Stephen Trowbridge writes:
> Curtis,
> OK, so it is your place to tell SPs how to bill for their services rather
> than mine.

No.  It is the SPs business to tell us.  

> No doubt that there are certain services at certain usage levels where it
> is better for the SP and for the customer to bill at a flat rate. All I am
> saying is that this is not necessarily the correct model in all cases.
> Sure, I can pay a flat rate for ulimited long distance calling if I choose
> to, but this flat rate is about 10x my normal long distance bill - so I
> choose to pay per call instead. But if I follow your model, I can expect
> what I pay for very few long distance calls to increase dramatically in
> the near future, since nobody will be able to measure my usage for VoIP
> and their only choice will be to charge me the amount that covers average
> usage.

No SPs seem to be making credible requests for customer initiated MPLS
connections which are per bandwidth billable.

> I pay a flat rate (~$45/mo with taxes) for my cable modem service. This is
> a bargain since we have 4 computers (2 tech. industry professionals, 1 CS
> major college student, one high school student) behind my router and we
> make pretty heavy use of it. But would you blame a SP for wanting to provide 
> a
> metered service also to try to get more cable modem penetration among light
> users?

But they don't and they don't intend to.  AOL could also go back to
metered billing.  So could the cell phone providers.  Customers
disappear, so they don't meter.

> Again, these are market related decisions to be made by the SPs. Not
> technology related decisions to be made by the equipment designers.
> Regards,
> Steve

The discussion was about requirements and I am speaking on the basis
of my experience on the SP side and what requirements I have heard on
the vendor side and observations about very clear industry trends
(toward not away from flat rate billing).

Nice tactic though.  After you lose the argument, declare the argument
out of scope.  :-)

Curtis


> Curtis Villamizar wrote:
> > 
> > In message <3E64EAA2.FDD36434@lucent.com>, Stephen Trowbridge writes:
> > > Curtis,
> > > Working for an equipment manufacturer, I have never felt that it was
> > > my place to tell a SP how they should bill for their services. I don't
> > > think we should preclude any particular billing model by our architecture
> .
> > > It is the market that should decide what gets offered at a flat rate:
> > > not the equipment, and certainly not the standards.
> > > Steve
> > 
> > Steve,
> > 
> > This is no different from the 1995 argument that IP should go away
> > becasue there is no good way to bill on a per connection basis.
> > 
> > What you are asking for is customer initiated connections which can be
> > metered for billing purposes.
> > 
> > I used to work at two major ISPs in the architecture group.  There was
> > no such requirement.  Since leaving and being on the vendor side I
> > have not heard any such requirement from customers.
> > 
> > Well actually if you count Enron as a customer, Enron's bandwidth
> > brokering required customer initiated connections with AAA and
> > possibly metering.  They hadn't decided on the latter before they gave
> > up and switched their business plan to a QoS enabled backbone for VOD
> > (remember the Enron/Blockbuster announcement).  We all know how far
> > both of those ideas went.
> > 
> > It appears that you are not arguing that any SP wishes to repeat
> > Enron's folly (the first of the two), just that we should extend the
> > protocols to accommodate it just in case.  That doesn't sound like a
> > strong case for a set of requirements.
> > 
> > Curtis
> > 
> > > Curtis Villamizar wrote:
> > > >
> > > > In message <3E64A690.D3BB8603@lucent.com>, Yangguang Xu writes:
> > > > >
> > > > >
> > > > > Kireeti,
> > > > >
> > > > > Yea, you are right. You pay flat rate for a DSL.  [...]
> > > >
> > > > Lets look up and down the food chain.
> > > >
> > > >   At consumer level, service is tiered flat rate.  DSL, cable IP, cell
> > > >   phone, now POTS.
> > > >
> > > >   IP service at T1 or higher has always been flat rate or burstable
> > > >   tiered flat rate.
> > > >
> > > >   Switched services such as FR and ATM have been flat rate PVC or
> > > >   SPVC.  SVC service is generally not available and where it is has
> > > >   almost negligible penetration.
> > > >
> > > >   At the high end, leased lines, leased lambdas, and leased dark fiber
> > > >   is definitely not signaled dynamicly by the customer so there is not
> > > >   "call" or customer connection.
> > > >
> > > > What has had a call model:
> > > >
> > > >   X.25
> > > >
> > > >   ISDN
> > > >
> > > > Even ISDN, where it has been at all successful, is flat rate.  Where
> > > > it is metered, market penetration is zip.  (btw - Its still metered
> > > > where I live afaik and might as well not be offered).
> > > >
> > > > > Furthermore, you are talking about billing. Indeed, one of the reason
> s op
> > > erto
> > > > > rs
> > > > > choose flat rate is to simplify billing because the control/managemen
> t pl
> > > ane
> > > > > can't generate accurate information.
> > > > >
> > > > > The bottom line is that as the building block of the overall manageme
> nt
> > > > > functions, control plane should provide adequate function to support 
> high
> > >  lev
> > > > > el
> > > > > services. Otherwise, everything becomes commodity.
> > > >
> > > > So what does this billing and control/management plane nonsense mean
> > > > in a world with flat rate or tiered service with no customer initiated
> > > > connections to bill for?  It has absolutely no purpose except to try
> > > > to perpetuate a mindset and perpetuate a hope that services with the
> > > > X.25 and ISDN metered billing model will somehow gain penetration and
> > > > the SP will be able to milk huge profits from it.  Unfortunately the
> > > > customer isn't stupid enough, so short of a return to monopoly rule or
> > > > other exclusion of competition, it won't happen.
> > > >
> > > > > Yangguang
> > > >
> > > > Curtis
> > >
>