August 2009 - nml-wg - lists.ogf.org

Notes from Aug 27th Teleconference
by Jeroen van der Ham 27 Aug '09

27 Aug '09

Agenda: - Update of the schema document status - conversation with ITU - Status of Deliverable #1 and namespace request - Ports with multiple connections - NSI - Topics for OGF 27 Attending: Chin Guok Guy Roberts Aaron Brown Freek Dijkstra Paola Grosso Marcin Wolski Jeroen van der Ham Yufeng Xin Schema document: Not much happened outside of the schema section. No response from Lars yet, Freek is still working on use-cases. Jeroen submitted a new revision, and will continue to work on it. Comments on the Service: Aaron: The service element should remain generic, adaptations and switching are example instances Freek: Service should not be limited to Node. It will be changed to group. Freek: Does a service describe capabilities or also configuration? It was decided that a service description only describes capabilities. Actual configuration will be describe in a different way. Conversation with the ITU: John, Guy and Jeroen had a talk with some people from Ciena about the ITU standards. These people have been heavily involved in the ITU docs and provided some good insights. Notes about this call will be forwarded soon. They have volunteered to answer more of our questions. So if you have any after reading the notes, Jeroen can forward questions to them. Jeroen will also ask them to come to the next OGF meeting. Deliverable 1 and Namespace status: Freek: no change. Multiple connections on a single port: On the mailinglist we have had a discussion about a use-case where there is a single port with separate vlan sets that are split outside of the domain. At the end of the discussion we agreed that the external party that does the switching should be modelled as a static configuration. This brought up the question who should be responsible for topology descriptions. It was decided that this question is out of scope for NML. Freek suggested that we turn this into a use-case. Topics for OGF 27: Freek requested agenda items for OGF27. Jeroen volunteered to present the status of the schema deliverable. We had a discussion about the way to move forward right now and how to best organise the meetings at OGF. It was decided that we will have a somewhat regulated discussion. Freek will pick some issues that are still open on the mailinglist and put them on the agenda. The idea is that people prepare their points for these items so we can have a good discussion in the meeting. As for the decision process, the schema document should now be the authoritative source of definitions. Any problems or issues with these definitions should be sent to the mailing list or discussed at the OGF, along with a proposed change or addition to the document. Yufeng proposed that as an exercise people should try to model their network using the current schema. He will try this on their network and present his findings at the next OGF. Freek will examine the current issues on the mailinglist and put together an agenda for the meeting. Next call: Is scheduled in two weeks, September 10th, at the same time (14:00 UTC, 16:00 CEST, 10:00 EST, 7:00 PST).

1 0

OGF 27 sessions
by Freek Dijkstra 27 Aug '09

27 Aug '09

Hi all, On the call we proposed that we would make an agenda for Banff OGF with the following topics: - schema update presentation by Jeroen van der Ham - 2 or 3 topics pre-selected to discuss I just noticed that the call for sessions is already open at the OGF website (despite that I have not seen an announcement on the ogf-all list). Two questions: - How many sessions do we want? I propose 3 sessions of 90 minutes each. I like to make an initial request tomorrow. - What topics do we like to discuss? On the mailing lists there was a suggestion to discussion service, perhaps in two separate discussion (concept + definition and implementation), and to discuss one or two of the use cases. Regards, Freek Dijkstra

1 0

Thu, Aug 27, 14:00 UCT conference call
by Freek Dijkstra 27 Aug '09

27 Aug '09

Next NML conference call: Thu 27 Aug, same time (14:00 UTC, 16:00 CEST, 10:00 EST, 7:00 PST) Duration: 1:30 (copy from previous settings, I presume these are still correct). To join both the web and audio conference (recommended), click here: https://edial.internet2.edu/call/0186145 If you only want to participate in the audio conference, you may dial in with the following numbers: Call: +1-734-615-7474 (PSTN CALL-OUT DOES NOT WORK) or +1-866-411-0013 (toll free US/Canada Only) Enter access code: 0186145 Agenda: - Update of the schema document (Jeroen) - Topics for OGF 27 (Freek) Regards, Freek Dijkstra

4 5

Ports with multiple connections?
by Jeroen van der Ham 27 Aug '09

27 Aug '09

Hello, I have been examing some of the topologies of the IDC network at Internet2 and others more closely and I have encountered some things that we may want to look at. One of the things that I would not know how to describe exactly in NML is the following example: A domain has a port on one of its devices. This port has an inter-domain connection. Through this port there are two possible connections to other domains, by means of separate VLAN labels. For example VLANs 1000-2000 go to one domain and 3000-4000 go to another. There are different ways of solving this at the moment. You do have to describe this somewhere as a single port, because of the limitations on the capacity of this port. I came up with three possible solutions, all with their own disadvantages: - Describe a single port with two links attached to it. This means breaks the assumption that a port can have at most one bi-directional link attached to it. - Describe a single port with two virtual ports each with its separate link. This breaks the assumption that a port is always connected by a Link (either a cross-connect or an external link), and we introduce a new concept of a virtual port. Are there other cases that would warrant this introduction? - Describe a single port that is connected to a "virtual" node where the connection is split into two. This makes the description more complex, and also seems to allow a direct connection between the two inter-domain connections without using the original port. Any other ideas? Jeroen.

3 9

Agenda for OGF 27
by Freek Dijkstra 25 Aug '09

25 Aug '09

I propose to discuss the OGF 27 agenda on the call next Thursday. How do we want to organize the meeting? General talks, discussion on the schema? Should we set topics beforehand, and if so, which ones? The discussion in the last two months have been good, but there are many more things to discuss before we have general consensus on the schema. Regards, Freek Dijkstra

1 0

New version of the schema document
by Jeroen van der Ham 24 Aug '09

24 Aug '09

Hello, I've just submitted a new revision of the schema document to the subversion repository. Here's the changelog: - Added a short description of the Location and Lifetime objects - Added a section enumerating all Relations. I've not called them "Relations" throughout the text. If need be, we can always make the names of Relations implicit in the syntax. - Added a section on Service, describing the way we currently see adaptations? I welcome comments! - The group section probably needs more work. I've updated the UML schema diagram to the latest version. I've also added a note saying that this is not intended as endorsement or finalizing, merely trying to reflect the current status. - Changed the Link group to Bidirectional Link as per the diagram. The latest version of the PDF can be found in the subversion repository. Unfortunately the web access to it seems broken at the moment, so I'm attaching it to this mail. Jeroen.

2 2

Terminology
by Jeroen van der Ham 20 Aug '09

20 Aug '09

Hello, The NSI group has also been studying the G.800 terminology, and has attempted to make a table with the different terms in NSI, G.800 and NML. NSI G.800 NML --- ----- --- Layer Topology Layer Network N/A[1] Subnetwork Subnetwork Network/SwitchMatrix[2] Link Link Link[3] Port Link Point Port Forwarding Port Forwarding Point Port Segment Link connection Link Subnet connection Segment Access Port Access Group Port?[4] [1]: The Layer topology concept is not something that we describe explicitly. Nor do I think that we really need a name for it. It describes a group of nodes on the same layer. We can describe nodes, and are planning to describe what layer nodes are on. Getting this group of nodes is then just a simple query. [2]: A Subnetwork in G.800 terminology is the place where the switching occurs. However, the subnetwork concept is used at different abstraction levels, and we have not decided yet how to represent different abstraction levels. [3]: The Link concept is still somewhat vague. G.800 uses the term Link to describe a connection between two ports (G.800: points). However, they also describe a concept of multiplexing using forwarding ports, creating a bundle of links on the client layer, this bundle is not explicitly named in G.800. I'm not entirely sure whether NSI Link concept is the NML Link concept, or the G.800 bundle (or both). [4]: Access Port is the point at which a client connects to the network. In my mind NML does not need to make a special distinction for this, and we can just call it a Port. Jeroen.

2 1

Notes of conf call Aug 14
by Freek Dijkstra 14 Aug '09

14 Aug '09

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)

1 0

Ethernet use case for NML
by Guy Roberts 14 Aug '09

14 Aug '09

Hi all, I have come up with a diagrammatic version of Freek's Ethernet use case. Perhaps we can discuss on this call? Guy ------------------------------------------------------------------------------------------------------------------ Guy Roberts, Ph.D Network Engineering & Planning DANTE - www.dante.net<http://www.dante.net/> Tel: +44 (0)1223 371 316 City House, 126-130 Hills Road Cambridge, CB2 1PQ, UK ------------------------------------------------------------------------------------------------------------------

1 0

New document
by Freek Dijkstra 14 Aug '09

14 Aug '09

To top it off, I just noticed that Martin committed a (minor) revision of deliverable #2 this morning (it's rather annoying that subversion does not sent mails out, I will look into this). To facilitate people without subversion access, I'll attach the document. For those who want check out the subversion repository: svn checkout http://forge.ogf.org/svn/repos/nml-base Regards, Freek

1 0