Re: [PCN] about the signalling draft

[Fortune HUANG <fqhuang at huawei.com>]

Hi Lars,

I was not suggesting specifying any specific signalling protocols, either.
What I meant is the signalling requirements for the two logical interfaces.

Best regards,
Fortune


-----Original Message-----
From: Lars Eggert [mailto:lars.eggert at nokia.com] 
Sent: Tuesday, November 10, 2009 10:38 AM
To: Fortune HUANG
Cc: pcn at ietf.org
Subject: Re: [PCN] about the signalling draft

Hi,

On 2009-11-10, at 10:22, Fortune HUANG wrote:
> As pointed out by Lars just now in the meeting room, the signalling
requirement draft should be generized to be applied to all deployment
scenarios or implementations.
> I think what we actually need to standardize in the signalling draft is
two logical interfaces: 1) the interface between the PCN-ingress-node and
the decision node; 2) the interface between the PCN-egress-node and the
decision node.

to be clear: I was taking about capturing *requirements* for PCN signaling
protocols, which is what the ID tries to do.

I was *not* arguing that we should standardize any specific signaling
protocols, for either 1 or 2 or cases with no central decision node.

Lars

> We could also mention the common deployment scenarios in the draft (but
please note that this is not the intention of the draft and this could even
be put into the appendix), e.g.:
> 1) Scenario 1, the decision node co-locates with the PCN-ingress-node:
the interface between the PCN-ingress-node and the decision node is an
internal interface which can be implemented privately within a vendor; the
interface between the PCN-egress-node and the decision node is an external
interface which must follow the standardized interface.
> 2) Scenario 2, the decision node co-locates with the PCN-egress-node: the
interface between the PCN-ingress-node and the decision node is an external
interface which must follow the standardized interface; the interface
between the PCN-egress-node and the decision node is an internal interface
which can be implemented privately within a vendor.
> 3) Scenario 3, the decision node co-locates with neither the
PCN-egress-node nor the PCN-ingress-node: both the interface between the
PCN-ingress-node and the decision node, and the interface the
PCN-egress-node and the decision node are external interfaces and must
follow the standardized interfaces.
> 
> There was an argument whether the admit/block state or the raw data could
be sent from the egress to the ingress. From my understanding, this argument
relates to two deployment scenarios: when reporting the admit/block state,
we are refering to the interface between the decision node and the
PCN-ingress-node, where the decision node is co-located with the
PCN-egress-node; when reporting the raw data, we are refering to the
interface between the  PCN-egress-node and the decsion node, where the
decision node is co-located with the PCN-ingress-node. So both reportings
could co-exist since they belong to different interfaces. Since both
scenarios are valid, we don't need to make a choice at least in the
signalling draft.
> 
> Best regards,
> Fortune
> 
> <ATT00001..txt>