Re: [GLIF all] proposal for the introduction of Global Identifiers for lightpaths

[Ronald van der Pol <rvdp@xxxxxxx>]

On Fri, Aug 22, 2008 at 11:18:22 -0400, Mathieu Lemay wrote:

> I find that scheme and host location to be important. When using URIs you simply have
> URN as the scheme when doing management display or transient service (Resource ID), however it is inadequate to access directly the resource...
> 
> What that means is that the same resource may have the following access mecanisms in different contexts:
> 
> 1) urn:glif:northernlight:G6O76GQ 
> - Here this kind of ID are useful when looking at "indices" in a resource repository. However it gives the application no idea on HOW /WHERE this resource can be accessed. So this cannot be a service / resource endpoint but can be part of the datamodels.
> 
> 2) x-jms://northernlight:xxxx/resources/lightpaths/glif/northernlight/G6O76GQ
> - This would be the equivalent "routing" for messages going through JMS for the same resource.
> 
> 3) http://www.northernlightserver.com/resources/lightpaths/glif/northernlight/G6O76GQ
> - This would be the RESTful address to this resource or could also stand as the WS-Addressing Endpoint for SOAP services...
> 
> Notice that URIs can also have fragments thus:
> 
> http://www.northernlightserver.com/resources/opticalswitch/glif/canarie/ome-ott01#CardX

Mathieu,

I do not think we should overload the semantics of the global identifiers.
I think it is useful for operating lightpaths to have the sourcing
organisation the the identifier. That is the party who coordinated the
setup of the lightpath and probably has the most knowledge about the
lightpath.

Having addintional services is useful. But what if the sourcing organisation
wants to subcontract the running of these services to another party/domain?
And what if they change subcontractor? Does the global identifier change?
That does not sound right to me.

	rvdp

Attachment: signature.asc
Description: Digital signature