[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: DS-TE Requirements - input sought on "Overbooking"



On Mon, 5 Nov 2001, Francois Le Faucheur wrote:

> To all Network Operators interested in DS-TE,
>
> > > Do we agree it is worth adding something on Overbooking in the "DS-TE
> > > Reqts" draft?
> >
> >The concept of overbooking should be addressed in the requirements
> >draft. However, i am not sure if overbooking is necessary in the
> >protocol extensions.  There are alternative ways to achieve over-
> >booking, including adjusting bandwidth of particular LSPs.  While
> >this doesn't allow you to overbook different parts of the network,
> >i question whether adding this justifies the increased complexity.
>
> Fair question. So, let's poll for an answer before we put it in the
> requirements draft:
>
> Considering a DS-TE solution which allows to enforce different overbooking
> ratios for different Classes of Service, are there Network Operators out
> there (I repeat *Network Operators*) who have a specific requirement for
> enforcing different values for these per-COS overbooking ratios in
> different parts of the network ?

and as a question of delineation....

is this a function of protocol, or back-end configuration systems?

back end systems, which can be used to place a bandwidth on an LSP take in
a series of values V[1:N] and run them through a function to distill a
scalar.  This function may take in the class of the LSP as an input, and
could use a class specific oversubscription rate, and can do things no
protocol can do. As an example....  say you have the following:

	V	time series of bytes on an LSP
	dV	differential of V (e.g. dV[i] = dV[i] - dV[i-1] / time)
	gdV = g(dV)
		scalar gdV returned by function g
	bw = f(gdV)
		a non-linear mapping of the scalar gdV

f(gdV) may allow for more overbooking of smaller LSPs than of larger ones.

Also, links may also be adjusted to allow for overbookings as well.

So, Network Operators - to rephrase the question:

Is there a need for per-COS and per-Link based overbooking which requires
protocol hooks for advertising out per-COS, per-Link overbooking factors?

My answer is no.



 > > > Thanks
>
> Francois
>
> >-dave
> >
> >
> > >
> > > Are the two points above reasonable?
> > >
> > > Thanks for your thoughts
> > >
> > > Francois
> > > _________________________________________________________
> > > Francois Le Faucheur
> > > Development Engineer, IOS Layer 3 Services
> > > Cisco Systems
> > > Office Phone:          +33 4 97 23 26 19
> > > Mobile :               +33 6 89 108 159
> > > Fax:                   +33 4 97 23 26 26
> > > Email:                 flefauch@cisco.com
> > > _________________________________________________________
> > > Cisco Systems
> > > Domaine Green Side
> > > 400, Avenue de Roumanille
> > > 06 410  Biot - Sophia Antipolis
> > > FRANCE
> > > _________________________________________________________
> > >
>
>
>