[NM-WG] Meeting notes

[Freek Dijkstra]

These are the draft version of the meeting notes for the June 4, 2008 
meeting at OGF 23.

Please give comments and corrections.

About 10 attendees are present.

Relation with other working groups
==================================

On Monday, Martin presented a BOF for a Network Measurement [and] 
Control (NMC) group. At the NMC meeting, some people thought that some 
of the work items for the NMC group were already in the charter for the 
NM-WG.

The consensus is that the NM-WG should complete the "schemata doc".
The NMC group should document extensions to the schemata doc.

The NML working group is making progress, but it was expected that a 
final standard from that group would still be a year away.

Discussion of the V2 & V3 & NML Topology schemata
=================================================

Martin Swany looked at the impact op the deployment of the three 
topology schemata. There are incompatibilities in the current installed 
base, and it is agreed that one version should be recommended. Martin 
Swany and Aaron Brown recommend to move forward to version 3 of the schema.

There are a few changes in address elements, host pairs and namespace, 
but these should be straightforward.

Martin Swany says that v3 of the NM schema is nearly exactly the same as 
the NML schema to be. Freek Dijkstra says that even so, he expects that 
it is not exactly the same to the letter. Richard Hugh-Jones says that 
by the time NML comes out, there will be a large installed base, and 
converting than is not easy. Cees says that NML will have a new features 
that should act as a carrot for implementers.

Freek Dijkstra notes that waiting for NML is not an option at this 
moment, so it is a choice between v2 or v3. It is decided to only have 
version 3 in the documents, and recommend implementers to follow that 
version.

Yuri Demchenko had a remark in the NML group that the OGF discourage the 
use of URNs, in favour of URIs (so http://... instead of urn:...). 
Martin Swany nor Richard High-Jones ever heard about it, but Richard 
will check with the steering committee. Martin remarks that the schema 
does use URIs for the namespace; only the resource identifiers use URNs.

Administrative Actions
======================

The slide of Richard Hugh-Jones says that the schemata document should 
include information about "protocol, SOAP, and document literal 
information". Freek Dijkstra ask what is meant with this. Richard 
clarifies that the document should specify how the XML data is wrapped 
in a message. It should not contain a state machine. That is a taks for 
the new NMC group, if it is chartered.

Jeroen van der Ham suggests to take the topology specific parts out of 
the document and put it in a separate document. This makes it easier to 
later replace only that part with the NML schema, without altering the 
rest of the document.

The current document still contains v2 examples. It is decided to 
replace them with v3 example rather than marking them as obsolete.

Freek Dijkstra suggests to add an appendix on what's changes from v2 to 
v3. This will make it easier for implementers to make the changes.

Discussion of the V2 and V3 schemata
====================================

Freek Dijkstra asks about the current schema, and there is a short 
discussion on unidirectional and bidirectional links, how link 
information is fetched in NDL (manually or automatically). Links are 
unidirectional, and that will be given as input to the NML working group.

Tasks
=====

- Martin Swany will rewrite document
   - the version 2 examples are replaced with version 3 examples
   - it is recommended to put the topology specific sections in a 
separate documents
   - it is recommended to add an appendix highlighting the difference 
between v2 and v3 schema
- Richard Hugh-Jones asks the GFSG is there is a policy on the use of 
URI and URN, and if that applies to schemas or to resource identifiers.