[Nsi-wg] minutes from today's NSI call

[Guy Roberts]

The minutes from today NSI conf call are attached below.
Guy
-------------
Minutes NSI-WG conf call 15 Sept 2010



Attendees:

Guy Roberts

John Vollbrecht

Chin Guok

Tomohiro Kudoh

Jerry Sobieski

Inder Monga



Apologies:

Radek Krzywania

Joan Garcia-Espin


Action points from last week's call:

AP1: JS to prepare a brief proposal for event notifications in the Connection Service - ongoing

Discussions:

Agreement that the 'modify' primitive will not be included in version 1.0 of the Connection Service

The 'query' primitive is to be included.  In its basic version this primitive will allow the requester NSA to request the state of any attribute for a given connection Id.  More advance queries are still for discussion.

Discussion on how to authorize primitives.  General agreement that each primitive should include its own authorization attributes.  No agreement on how these should be used.

Discussion on the problem of keeping Connection state in NSAs.  This could cause scaling problems (i.e not scalable) since each NSA needs to keep the state of each transited connection and keep track of who requested the connection and who it was forwarded to.  This issue needs to be analysed further in future once the protocol has been defined.

Notifications will be included in release 1.0.  No agreement on what can be notified and how notifications are enabled.
Agreement was reached on the principle that notifications should only be sent to the requestor NSA that requested the Connection.  If notifications are to be sent to other NSAs (not the Connection requester) there needs to be a mechanism to identify these to the provider NSA.  This mechanism has not been defined.

Discussion on whether we really need two mechanisms for provisioning - auto and manual.  Should we reconsider this decision and go for just one of these methods? Or should we stick to the current agreement as documented in the Connection Service Architecture document.

Jerry suggested revisiting the decision to define the 'start time' as the time at which provisioning is completed and the connection comes into service - is this realistic?  Can operators guarantee that we can meet this requirement?


Action Points:



AP1: GR to book 3 NSI sessions at OGF 30 (prefer Tuesday)

AP2: JS to prepare a brief proposal for event notifications in the Connection

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.ogf.org/pipermail/nsi-wg/attachments/20100915/4de3ac2e/attachment.html