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

Re: [ANCP] Question about the ANCP high-level architecture

stefaan.de_cnodder at alcatel.be wrote:


Hi Kais,

see below...

Kais Belgaied wrote:

Folks,

My question is regarding the following picture of the high level ANCP
architecture:


                                               +--------+
                                               | Policy |
                                               | Server |
                                               +--------+
                                                    |
                                                    |
  +-----+   +-----+   +--------+                 +-----+   +----------+
  | CPE |---| HGW |---|        |                 |     |   |          |
  +-----+   +-----+   | Access |   +---------+   |     |   | Regional |
                      |  Node  |---| Aggreg. |---| NAS |---| Network  |
  +-----+   +-----+   |        |   |  Node   |   |     |   |          |
  | CPE |---| HGW |---|        |   +---------+   |     |   |          |
  +-----+   +-----+   +--------+                 +-----+   +----------+


The policy server is attached to the NAS only.

A couple of observations about the AN:
. it is shared between multiple HGWs that don't know/trust each other
. it is the first candidate for enforcing link sharing policies (e.g. bandwidth
allocation) between subscribers attached to those HGWs


do you mean that bandwidth can be shared between 2 subscribers? This can also be done by the NAS Policy server I would


yes, that's what I mean. I understand the information about all subscribers is available at the Policy server.

think. Note that the main focus of the WG is DSL, and sharing bandwidth in DSL is rather difficult.


I'm possibly seeing more sharing in the aggregation node component than there actually is with DSL.
I view it as a resource shared between multiple subscribers, therefore the AN being the arbitrator for
accessing that resource would be a natural choice.
If in reality (and in the foreseeable future) the model for the aggregation node is just a collection of
independent pipes to to the NAS, then, yes the NAS is the first point shared access, and the policy
enforcement belongs there.


Such policy is applicable to HGWs connected to the same AN.
There may be other policy information still live in the NAS policy server.
The picture could look more like:

                       +--------+              +--------+
                       |   AN   |              |  NAS   |
                       | Policy |              | Policy |
                       | Server |              | Server |
                       +--------+              +--------+
                           |                        |
                           |                        |
  +-----+   +-----+   +--------+                 +-----+   +----------+
  | CPE |---| HGW |---|        |                 |     |   |          |
  +-----+   +-----+   | Access |   +---------+   |     |   | Regional |
                      |  Node  |---| Aggreg. |---| NAS |---| Network  |
  +-----+   +-----+   |        |   |  Node   |   |     |   |          |
  | CPE |---| HGW |---|        |   +---------+   |     |   |          |
  +-----+   +-----+   +--------+                 +-----+   +----------+

comments?


As mentioned during the meeting, the interaction with the policy server has to be further detailed. I do not say that the above architecture is invalid but I do not see how it relates to the current framework of ANCP. The NAS policy server gives all subscriber information to the NAS, which enforces the policies are gives them to the AN via ANCP. If you have 2 policy servers, then you must also take care of all possible interactions, e.g., AN policy server says that the bandwidth is x Mbps while the NAS says y Mbps.


I should've clarified: the AN policy database is a subset of the whole policy database residing in the NAS policy server.
the AN server caches in the information relevant to subscribers connecting through it *only*.
and yes, that'll add a requirement on the AN to keep the cached version in sync with the NAS',


Maybe an Element Management system to the AN can be added because that one has some interaction with ANCP. 


makes sense.

   Kais.


regards,
Stefaan



_______________________________________________
ANCP mailing list
ANCP at ietf.org
https://www1.ietf.org/mailman/listinfo/ancp