Attending:

Franco, Javier, Pete, me, Gigi, Afrodite (from EGEE)


5) Attendance for GGF11:
====================

Pete

Franco expects attendance to be fairly weak. Franco will not be in Hawaii.
Needs to know whether to have one or two sessions by the end of the week.

Franco: So one session should be enough. If there are meetings on the side, then that's good.

Pete: If things have gone far enough, can we have a cross-session with OGSA people.
The deadline for papers for GGF11 is May 14. Need to redo netservices document, as it doesn't really cover what we want already.

1) Use cases
====================

Pete: His draft: he just wanted to try to illustrate how a use case should work.
You need to write down what the client wants to do, and then try to write that down without writing down anything even close to a solution.

Pete has no objection to the OGSA template. Doesn't seem particularly business-like.
But he feels he should add a walkthrough to the middle of the headers there, which steps through every step that the client does.


Franco: For lightpaths:
What are the endpoints? Could be an IP address.


Franco: so Pete's and Gigi's use cases are quite similar.

Other use cases:

Pete and Javier have discussed about use cases. A few for HEP and some other things. But not use soon.

Javier: also requirements for operations in EGEE. Is that covered by GHPN?

Afrodite: There's a complex hierarchy in EGEE. Currently trying to decide which parts should be operations requirements, middleware, users.
Where in the hierarchy should operations be put.

Franco: what exactly is operations:
Afrodite: 

Afrodite: Within EGEE there is not just a single administrative domain to deal with.
How can network provisioning be integrated with the monitoring.

Franco: the path will be the end to end composition of all the path elements. But how are the path elements defined? Is it adminstrative steps or perhaps technological. To Franco, the path element is whatever the domain administrator puts into the database. It doesn't matter if there are a number of control planes between us and the network as long as there is an ingress point and an egress point.



Gigi: in Canarie, the users control individual hops on path elements.
Franco: however, *someone* is responsible ultimately. So there is still an administrative angle that can be made use of. Don't know how the model fits in, though.
Gigi: the problem is that it is not adminstrators who configures stuff, but it is signalling (i.e. user controlled).
However, the hardware is still maintained by some entity.



Franco: has been talking to a group (a company called Mitre, a research operation for the US Air Force) who have a very large hierarchy of attributes related to network services.
Of course, the GGF boilerplate has to be attached.
We will probably get to look at that in a couple of weeks.

There is also CIM, but this is limited to layer 3, and has nothing to do with services.

If anyone knows of any further similar sources for network elements and paths.

AOB? Nope.

So, there should be more work on use-cases. We adjourn until next monday the 10th, when we have our next meeting. We will also have better defined plans for what to submit for GGF11.


After:

Some books:
Writing effective use cases. Cockburn.