Peoples, As promised I have done a first pass at modelling the additional NSI topology concepts in NML. I did an overview of all the components we have defined to date so we can see how they all hang together. The slide pack attached is an overview of the proposal. The XML document attacked is an updated reference topology representation, and the schema files attached are the updated NML, NSI-EXT, and service definition schemas. John
Hi John, Clear proposal, thanks! I am halfway reading this, but have a different meeting coming up, so I just comment on what I read. (slide 4)
– An AdaptaNon can also be defined between STP of the same service type in the case where encapsulation/adaptation of the input service type results in the same output service type.
I'm not clear on the purpose of this type of adaptation (one type encapsulated in the same type). Can you perhaps give a example? (slide 5) What is the purpose of having different Service Domains in a Network Topology? Because the STP in each services domain are of a different type? Or to describe geographic subdivisions withing a Network Topology? Note that the former is represented in NML using multiple SwitchingServices, and the later is represented in NML using multiple (sub) Topologies 'in' a larger Topology (where this 'in' is formally defined as a 'hasTopology' relation). PS: I now see that you answered this on slide 19. Thanks, John! I love this proposal. My head is too blurred to due lack of sleep to really grok every character in the example, but from a first look, it looks very, very good! Freek On 04-12-2013 06:06, John MacAuley wrote:
Peoples,
As promised I have done a first pass at modelling the additional NSI topology concepts in NML. I did an overview of all the components we have defined to date so we can see how they all hang together. The slide pack attached is an overview of the proposal. The XML document attacked is an updated reference topology representation, and the schema files attached are the updated NML, NSI-EXT, and service definition schemas.
John
_______________________________________________ nsi-wg mailing list nsi-wg@ogf.org https://www.ogf.org/mailman/listinfo/nsi-wg
-- Freek Dijkstra | Group Leader & Network Expert | Infrastructure Services | SURFsara | | Science Park 140 | 1098 XG Amsterdam | +31 6 4484 7459 | | Freek.Dijkstra@surfsara.nl | www.surfsara.nl | Available on Mon | Tue | Thu | Fri |
Comments in line. On 2013-12-04, at 6:55 AM, Freek Dijkstra <Freek.Dijkstra@surfsara.nl> wrote:
I'm not clear on the purpose of this type of adaptation (one type encapsulated in the same type). Can you perhaps give a example?
Jerry brought this specific Ethernet example up, and I read through the standards to verify. The Q-in-Q double label is a single instance of the more generic 2..N label stacking supported by the standards. Jerry has the requirement to allow for an arbitrarily number of labels to be pushed and popped as the Ethernet packet transits the network. In this specific example, the packet arrives on one port in the domain, is a label is added, and it leaves another port in the domain with the same encoding. Somewhere else in the network the labels will be popped an equivalent number of times.
(slide 5) What is the purpose of having different Service Domains in a Network Topology? Because the STP in each services domain are of a different type? Or to describe geographic subdivisions withing a Network Topology?
The definition of a Service Domain is that it contains STPs that can be connected without restriction. When STP are of different types then they must be in different service domains. In fact, if a network does not support label swapping then each STP with the same label value will be grouped in their own Service Domain.
Note that the former is represented in NML using multiple SwitchingServices, and the later is represented in NML using multiple (sub) Topologies 'in' a larger Topology (where this 'in' is formally defined as a 'hasTopology' relation).
I understand the distinction.
PS: I now see that you answered this on slide 19. Thanks, John! I love this proposal. My head is too blurred to due lack of sleep to really grok every character in the example, but from a first look, it looks very, very good!
Thank you kind sir!
participants (2)
-
Freek Dijkstra -
John MacAuley