Re: SLA & admission control in DiffServ

Would you sign a contract to guarantee bandwidth on a variable
bandwidth network? I wouldn't, so what you are describing is
an SLA that should never exist.

I don't know what a dynamic SLA is. We certainly don't have any
standards for such things.

You might well have rate control in a source, admission control
in an ingress router, and policing in a downstream router.
I suggest you read RFC 3086 and RFC 3290.

   Brian

Amoakoh Gyasi-Agyei wrote:
> 
> Hi,
> 
> Should the SLA be static or dynamically negotiated? Assuming a static SLA with bandwidth limit, how can the service provider guarantee the
> SLA on a connection with a wireless component which is spatio-temporarily varying and whose bandwidth can degrade ungraciously? What is
> the need to differentiate a policer from an admission controller?
> 
> So, Brian, do you mean that the admission controller in this case will operate in the traffic source while the policer sits somewhere else
> or be integrated? I just do not know why and how some of these functionalities were separated although their functionalities appear to be
> very similar, if not identical.
> 
> Thanks,
> Amoakoh
> 
> >
> > Message: 3
> > Date: Wed, 27 Nov 2002 17:17:57 +0100
> > From: Brian E Carpenter <[email protected]>
> > Organization: IBM
> > To: Geunhyung Kim <[email protected]>
> > Cc: diffserv <[email protected]>
> > Subject: Re: [Diffserv] SLA & admission control
> >
> > It depends on the PDB in use. If it has a bandwidth limit,
> > then ingress policing will be used to discard excess traffic.
> > In this case, it may be desirable to apply admission control,
> > rather than let the policing throw traffic away. Actually,
> > that is something that the SLA should specify.
> >
> >      Brian
> >
> > Geunhyung Kim wrote:
> > >
> > > Hello, All,
> > >
> > > I have some confusion about the SLA and admission control.
> > > When the provider contracts the SLA with the customer
> > >  - the provide check if the network can support the SLA and the provider make the contract when the SLA can be supported.
> > >
> > > As far as I know, when the SLA mechanism is used in the DiffServ domain, the admission control mechanism is not required, I thought.
> > > Is my thought correct ?
> > > If not, please let me know why the admission control is required after SLA is entered into agreement.
> > >
> > > Thanks in advance,
> > >
> > > Geunhyung Kim
> > >
> > > None of us is as smart as all of us
> > > ==========================================
> > > Geunhyung Kim
> > >
> > > E-mail: [email protected]
> > >
> > > Tel: +82-54-279-5655
> > > Fax: +82-54-279-5699
> > >
> > > Networking & Distributed Systems Lab.
> > > CSE
> > > POSTECH
> >
> 
> _______________________________________________________
> Amoakoh Gyasi-Agyei
> Centre for Internet Technology Research (CITR)
> Electrical & Electronic Engineering Department
> Adelaide University
> Tel: +61 8 8303 6209 (office)
> Tel: +61 402 638 141 (mbl)
> Fax: +61 8 8303 4405
> http://aaron.eleceng.adelaide.edu.au/Personal/amoakoh/

-- 
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Brian E Carpenter 
Distinguished Engineer, Internet Standards & Technology, IBM 
On assignment at the IBM Zurich Laboratory, Switzerland
_______________________________________________
diffserv mailing list
[email protected]
https://www1.ietf.org/mailman/listinfo/diffserv
Archive: http://www.ietf.org/mail-archive/working-groups/diffserv/current/maillist.html