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

Re: Stepping back from the ASON Routing Discussion



don

Don Fedyk wrote:

Hi Adrian

I had to step away from my computer for a snowstorm looks like there was a
bit of a flurry here too;-) I'll try to stick to the questions and answers as I know them.


See inline


-----Original Message-----
From: Adrian Farrel [mailto:adrian@olddog.co.uk] Sent: Tuesday, March 16, 2004 5:44 PM
To: ccamp@ops.ietf.org
Subject: Stepping back from the ASON Routing Discussion



Just stepping back a moment...


It feels to me that we are getting drawn into discussions of what can and cannot be achieved using existing protocols. This is all very valid, but is clearly not part of the requirements work.

I understand that the DT wishes to analyse the changes to existing protocols to meet the requirements under the charter phrase "to capture what's missing from current CCAMP work", but I feel that this is clouding (and delaying) the production of a clear requirements draft.

So looking at the three topics that Deborah raised.

1.
Some members of the Design Team noted that reachability information (as described in Section 4.5.3) may be advertised as a set of UNI Transport Resource address prefixes (assigned and selected consistently in their applicability scope). These members of the Design Team raised a concern that existing methods of advertising reachability may need to be examined (on a per-protocol basis) to determine if they are also applicable for UNI Transport Resource addresses. They invite CCAMP discussion on this aspect.


The first sentence is about requirements.
Do we, or do we not, need to support advertising UNI Transport Resource address prefixes?

My opinion is Yes there is a requirement in signaling protocols requirement
to signal a name i.e. UNI TR address (complete).

i have some difficulties to map the above sentence with the below, for me reading this it is yes and then no i've another way so i don't need, and the below is the real problem of course the name resolution into the address prefix and one uses the address prefix not the name in the signaling, why shall you assume that the names are distributed but not their corresponding prefixes or the mechanism to resolve them?

No in inter domain routing. Requirement addresses must be able to be
aggregated and names are not. The requirement is that a UNI TR address must
resolve to an Prefix address (For GMPLS routing IP prefix fits). That leaves intra domain routing. The requirement is to have a destination
address resolution. You are signaling with a UNI TR address (complete) and
you have a IP address prefix or nothing. The problem is IP cannot take you
all the way because resolving completely outside of the area is not a
requirement. But inside the area you should be able to resolve a UNI TR
address to a IP address Control plane element.



The second sentence is about solutions and is not relevant in this draft.

2.
ASON does not restrict the architecture choices used, either a co-located architecture or a physically separated architecture may be used. Some members of the Design Team are concerned that GMPLS's concept of the LSR requires a 1-to-1 relationship between the transport plane entity and the control plane entity (Router). They invite CCAMP input on GMPLS capabilities to support multiple architectures i.e. how routing protocols would identify the transport node ID vs. the router or routing controller ID when scoping Link IDs in a link advertisement.


The first sentence is about requirements.
Do we, or do we not need to support a physically separated architecture with a 1-n relationship between control plane and transport plane entities?

I would say yes. The requirement I see here is devices not capable of participating fully in GMPLS routing.

but in 1-n, 1 controls n, and 1 participates to the routing, the devices (part of the set n) are data plane entities that are by definition not gmpls capable, so i don't understand what you try to say here by the above justification

The remaining text is about solutions and not relevant in this draft.

3.
In order to support operator assisted changes in the containment relationships of RAs, the GMPLS routing protocol is expected to support evolution in terms of number of hierarchical levels of RAs (adding and removing RAs at the top/bottom of the hierarchy), as well as aggregation and segmentation of RAs. These GMPLS routing capabilities are considered of lower priority as they are implementation specific and their method of support should be evaluated on per-protocol basis e.g. OSPF vs IS-IS. In addition, support of non-disruptive operations such as adding or removing a hierarchical level of RAs in or from the middle of the routing hierarchy are considered as the lowest priority requirements. Note also that the number of hierarchical levels to be supported is implementation specific, and reflects a containment relationship e.g. a RA insertion involves supporting a different routing protocol domain in a portion of the network.
Note: some members of the Design Team question if the ASON requirement for supporting architecture evolution is a requirement on the routing protocol (protocol specific
capability) vs. a capability to be provided by the architecture. For example, ASON allows for supporting multiple protocols within each RA. The multiple protocols share a common routing information database (RDB), and the RDB is the component, which needs to support architecture evolution. The Design Team invites CCAMP input to understand the protocol-specific impacts.


This seems to mix up requirements and solutions.
It is not relevant what OSPF and IS-IS can or cannot do.

The requirements questions are:
A. What does ASON require in terms of evolution of hierarchies? B. Are these requirements immediate and high priority?

No comment from me.




Is it reasonable to make this separation between requirements, and consequent required changes to the protocols? If so, can we focus on the requirements and make some rapid progress?


Hope that helps,
Don


Thanks,
Adrian






-- Papadimitriou Dimitri E-mail : dimitri.papadimitriou@alcatel.be E-mail : dpapadimitriou@psg.com Webpage: http://psg.com/~dpapadimitriou/ Address: Fr. Wellesplein 1, B-2018 Antwerpen, Belgium Phone : +32 3 240-8491