Global Lambda Integrated Facility

Re: please comment on global identifiers proposal

  • Subject: Re: please comment on global identifiers proposal
  • From: Ronald van der Pol <Ronald.vanderPol@xxxxxxxx>
  • Date: Wed, 6 Aug 2008 15:43:48 +0200

On Tue, Aug 05, 2008 at 13:01:44 -0400, Tom Lehman wrote:

> Ronald, All,
> 
> I have one question, hopefully a clarification.  
> 
> With the recommendation in section 4 to use the Sourcing GOLE Naming Scheme,
> the list of globablly unique GOLE names are as listed on the GLIF web site.
> 
> 
> My assumption is that if a circuit starts on a network which is not in that
> GOLE list, then the global id will still be based on the sourcing domain
> name.
> 
> For instance if we have circuit which has the following path:
> 
> Internet2DCN-MANLAN-NETHERLIGHT
> 
> Is it ok to use a name for instance:
> 
> dcn.internet2.edu-5705
> 
> since that represents the source domain?  
> 
> If so, I think we need to include in the recommendation, something to the
> effect:
> 
> "if the source domain is not in the GOLE name list, then globably unique
> names can be formulated by using the sourcing domain dns name information,
> as described in Section 3.2 for instance."
> 
> or should we just add the name "dcn.internet2.edu" to the list of GOLES
> listed on the GLIF web site?
> 
> Tom

Thanks for your reply.

This is something we need to discuss and get consensus about.

I was thinking that the first part is always a GOLE id.

Currently, many lightpath requests are still done by end-users by
sending emails to dozens of people and organisations. I think the
GLIF community would like end that practise and professionalise by
adopting the "sourcing organisation" setup procedure as mentioned
in the document:
http://www.glif.is/working-groups/tech/fault-resolution-0.9.pdf
I think this was decided at the interim meeting in Minneapolis.

Then the question becomes "who or what are the sourcing organisations".
I think I agree that it is more than just the GOLEs. It most probably
are also the NRENs like Internet2, Canarie, SURFnet, etc.

So, are we saying that the "sourcing organisation" should generate
a global identifier by:
- setting the first part of the global identifier to the identifier
  of that organisation; the organisation picks an id as long as it is
  globally unique.
- it sets the second part of the global identifier to some internally
  unique naming scheme
- the first and second part are separated by a hyphen.

Should we restict the global identifiers to alpha-numerics plus the
hyphen and have them case insensitive? This means "dcn.internet2.edu"
is not a valid string.

Do we allow multiple hypens in a global identifier? That would mean
that it is not possible to distinguish between first and second part
by just looking "left" and "right" of the hypen.

In your example could the global ID be something like:
INTERNET2-XXXXXX
or
INTERNET2DCN-XXXX
where XXXXXX is the second (local) part?

	rvdp

Attachment: signature.asc
Description: Digital signature