Notes of NML conference call Fri Aug 14, 2009 ============================================= Attendees: ---------- - Jeroen van der Ham - Guy Roberts - Martin Swany - John Vollbrecht - Freek Dijkstra - Chin Guok - Jason Zurawski (?) - Jeff Boote (?) Update schema text ------------------ Martin committed a minor revision of the deliverable #2 document: - most existing text intact. - only contains items for which there is some sort of consensus yet - removed switch matrix (no clear consensus) Martin says he agrees with Jeroen that "service" should be a very broad concept. Martin suggests to enumerate all possible relations Task: Jeroen will edit schema document and submit it before next call Lars and Freek have not done anything on identifier Jeroen will contact Lars TMF: ---- Freek doesn't think a formal relation is beneficial, given that TMF documents are not public. He encourages individuals to join the TMF, but doubts if that needs to be reported to the NML. Jeroen does not see a problem for the TMF to submit non-public documents to the NML. Freek thinks it is allowed according to the OGF policy, though it conflicts with the spirit of openness in the OGF policy. Sharing secret documents is not really possible, as everyone can join the NML mailing list or session. In the end it is entirely up to the TMF to decide what they want to share. ITU: ---- Freek: found list of authors of G.800: http://www.itu.int/net/ITU-T/lists/mgmt.aspx?Group=15&Period=13 Martin: why contact ITU? Jeroen: need to understand difference between Point and Port. Concepts is used a lot. Guy: NSI view is that we want to stick to G.800 terminology Martin: I thought NSI would stick to NML terminology Guy: yes. Jeroen: We want to submit Jeroen's questions to them. [PS: Freek: is this still necessary, as John explained the distinction between port and point later in the call. It seems to me that all Jeroen's question are answered.] ?: We can contact them and let them know we're working on this. Deliverable #1: --------------- Freek pushed it to the Area Directors. Taken to public comment as soon as they are back from holiday. NSI --- There is a NSI call next week with Victor Reijs on the stitching framework Discussion on examples: ----------------------- Jeroen: NML should be as complete as possible Martin: but it doesn't dictate everything Guy: Adding layers every time seems complicated Freek: Creating a new layer is not and should not be cumbersome Jeroen: G.800 explains a way to abstract away lower layers. NML should be able to do that too. Confusion: on the name "eth_utp_in" in Freek's example. Guy thought this means there is both a "eth_utp" and a different "eth_fibre" layer. That does not seem like a good idea. Freek explains this is not a layer. It is just an identifier, and could have been a random string. There are only three layers in the example: Ethernet, Fibre and UTP. Martin: Different modelling strategies. UNIS follows an object-based modelling strategy. Discussion on the distinction between point and port are John: ports are unconnected. Points are two ports connected together. (see figure 1 in G.800) E.g. port = end of a fiber port = begin of physical interface point = combination of end of fiber connected to begin of physical interface Martin: what does this distinction buy us? Everyone: nothing at all. Martin: Let's not clutter the NML with unnecessary G.800 stuff. Only need to include useful thing. [PS: Freek: NML is indeed not a G.800 reading group, though it seems like it lately.] John + Jeroen will follow up with diagrams of use cases as input to NML This is considered a reasonable approach. John: we need pictures with examples, and pictures should explicitly include layers. Layer/Technology Identifiers: ----------------------------- Freek: These are different identifiers from network element identifiers as we discussed half a year ago. Martin: use OGF namespace. Freek: We want it easy to extend. Allow everyone to invent a namespace. Martin: Use canonical namespace in OGF. And an additional organisation namespace, using the same syntax, but free to define by 3rd parties. Freek: What happens if two different organisations add an extension for the same technology, but a different namespace? John: Standardization is necessary for interoperability. Martin, Jeroen: This is exactly the same functionality as the enumeration for GMPLS by IANA. Also there there is need for standardization and a private space for non-standard usage. Next call: ---------- Thu 27 Aug, same time (14:00 UTC, 16:00 CEST, 10:00 EST, 7:00 PST)