Re: TML events: what events we need?

There may be some misunderstanding. 
> Again, the FE side TML implementation is a local matter.
> The reason I concentrated on the CE side is that an FE is not even 
> expected to have logic to process ForCES notifications.  It generates 
> such notifications, but does not receive them.  I would think it 
We havn't said that an FE should process ForCES (protocol) notifications. We never mentioned an FE would reveive any event from outer part. I'm not sure where comes such thought. 

We just mentioned that an FE TML is part of FE, therefore, events generated by a FE TML to FE PL is a process inside an FE.  Some events at FE side may trigger some FEO events and make report to CE. Where at the CE side, when CE TML report an event to CE PL, the CE PL may just process it by itself. There is no requirement and also it is not allowed that the CE PL reports the event to FE (we never mentioned we should do like this). 

Thanks,
Weiming

> inappropriate to require support for processing such messages as a 
> requirement on a local, internal, interface.
> 
> Yours,
> Joel
> 
> At 04:34 AM 8/16/2006, Wang,Weiming wrote:
> >We are discussing if the TML congestion event is necessary and how 
> >the event acts with PL layer. I just instanced it with how the event 
> >may act with FE side PL, to indicate the necessity. It is not hard 
> >to understand that if the event is necessary at FE side, regardless 
> >of how it is at CE side, we must define it.
> >
> >Still at the FE side, a TML is part of the FE. TML is not an 
> >independent realm. It is reasonable that FEO/FEPO LFB reflects 
> >something that happens at the TML. I have instanced the FE multicast 
> >list as an example, which is set by CE to FEPO and is further set to 
> >FE TML by the PL. Some TML events are the same, they are originated 
> >from FE TML, and trigger relavent FEO/FEPO events to CE.
> >
> >Thanks,
> >Weiming
> >
> >----- Original Message -----
> >From: "Joel M. Halpern" <[email protected]>
> >
> > > The local TML instance on the CE side should not be expected to
> > > generate LFB events.
> > > a) There is no suitable LFB Class to reference.
> > > b) There is no suitable LFB instance.
> > > The CD side TML should not be generating a PL event claiming that the
> > > event came from the FE Object or FE protocol object.
> > >
> > > There is no need for TML -> PL condition notifications to look
> > > anything like PL Event notifications.  They (local TML detected
> > > occurrances) are local notifications, and should be delivered using
> > > suitable local mechanisms.
> > >
> > > Yours,
> > > Joel M. Halpern
> > >
> > > At 07:20 PM 8/15/2006, Wang,Weiming wrote:
> > > >TML must be related to FE object/FE Protocol object LFBs. This does
> > > >not mean every TML event must triggire an FE object event, but some
> > > >TML events and some TML attributes  must be tied to the FE object
> > > >events and attributes. The attributes may include FE multicast list,
> > > >etc. That's not an implementation issue, I'm afraid.
> > > >
> > > >Thanks,
> > > >weiming
> > > >----- Original Message -----
> > > >From: "Joel M. Halpern" <[email protected]>
> > > >
> > > > > I do not think that we want to define an LFB class / instance and
> > > > > have the TML synthesize PL events.  The PL owns the ForCES
> > > > events, not the TML.
> > > > > I think that all we need to do is define the mandatory
> > > > > events.  Implementations may choose to implement additional ones.
> > > > >
> > > > > Yours,
> > > > > Joel
> > > > >
> > > > > At 01:51 AM 8/15/2006, Wang,Weiming wrote:
> > > > > >One general way is for the PL to report the TML event to CE 
> > (as a FEobject
> > > > > >event) by means of a highest priority. This presumes the CE-FE
> > > > control channel
> > > > > >is not totally blocked when the event happens.  CE will further give
> > > > > >a treatment
> > > > > >for the event, the treatment may include the reconfiguration of
> > > > some LFBs to
> > > > > >avoid such congestion.
> > > > > >
> > > > > >Another possible way is to let CE previously set some policy 
> > regarding the
> > > > > >treatment for the event when it happens. Tthe treatment may include
> > > > > >some action
> > > > > >like redirect data  rate limitation. This way is  to presume  the
> > > > > >total block of
> > > > > >CE-FE control channal when the event happens.
> > > > >
> > >