Re: globally unique identifiers for lightpaths?

[Akira Kato <kato@xxxxxxxxxx>]

I support the idea to give globally unique ID. I'd suggest the IDs
are not for gole but also "applicable lambda between GOLEs".

I'd propose to have IDs something like
	Lnnnn
where nnnn is a four digit number.

The description of Lnnnn can be posted to through web as well as
a DNS text record such as

$ORIGIN glif.is
L0011 IN TXT "netherlight:starlight:stm64 owned by xxx operated by yyy"

First token should be standarized within GLIF community so that we can
make a set of possible lightpaths mechanically even though the
info. NDL can also be easy -- just specifies the lamda ID. When the
port of the other end was changed, local description doesn't have to
be updated.  Just the other end operator is required update his/her
description.

-- Akira Kato


From: David Reese <dave@xxxxxxxxx>
Subject: Re: [GLIF tech] globally unique identifiers for lightpaths?
Date: Thu, 6 Dec 2007 09:31:41 -0800

> Ronald,
> 
> I support the idea of a global name/number.  I would propose that it  
> contain all of the local entity names and their local circuit names  
> which make up the global lightpath (rather than embedded in the name  
> per your last suggestion).
> 
> Dave Reese
> 
> On Dec 5, 2007, at 10:44 PM, Ronald van der Pol wrote:
> 
> > On Wed, Nov 21, 2007 at 14:42:20 +0100, Licia Florio wrote:
> >
> >> Dear all,
> >>
> >> Please find attached the minutes of the meeting held in Prague.
> >
> > Potential "Lightpath Management Group" members,
> >
> > One of the items that came up in Prague was global identifiers
> > for lightpaths. At the time I was not convinced this was needed.
> > That is beginning to change. The more lightpaths there are, the
> > more difficult it becomes to be sure that we are talking about
> > the same lightpath.
> >
> > One drawback that I still see is that it is an extra burden:
> > - for each new lightpath we need to come up with a name
> > - the name has to be put into the administration
> > - for those who want to keep local names the mapping
> >  between local name and global name needs to be put into
> >  the administration
> >
> > So, what do you think? Should we start using global names
> > for lightpaths?
> >
> > I think we should give it a try. If we agree we need a
> > naming convention. There are several possibilities.
> > Some examples:
> >
> > Just a unique ID consisting of a number.
> > pro: easy to generate
> > con: gives no semantical information at all
> > con: where do we keep a list of IDs that are already given out?
> >
> > end_site_1-end_site_2-number
> > pro: identies from where to where the lightpaths goes
> > con: no topological path information
> >
> > gole_1-gole2-....-gole_n-number
> > pro: topological information about the path
> > con: this can become a rather long string
> >
> > In the latter two cases the sourcing GOLE could give out the number.
> > This number has to be unique for that GOLE only (end_site_1 or
> > gole_1).
> >
> > Ideas? Comments?
> >
> > 	rvdp
> >
> >