Re: [i2rs] Call for Adoption by WG: draft-atlas-i2rs-architecture-01 (ends Aug 12)

[t.petch <[e-mail]>]

----- Original Message -----
From: "Alia Atlas" <[e-mail]>
To: "Joe Marcus Clarke" <[e-mail]>
Cc: <[e-mail]>
Sent: Tuesday, August 13, 2013 9:01 PM


> Hi Joe,
>
> Thanks for the detailed review and suggestions.  Responses are
in-line.
>
> Alia
>
> On Wed, Jul 31, 2013 at 6:57 AM, Joe Marcus Clarke
<[e-mail]>wrote:
>
<snip>
> > Section 6.4:
> >
> > Each I2RS Client will have an identity; it can also have secondary
> >    identities to be used for troubleshooting.
> >
> > JMC: Each application will have a _unique_ identity.
> >
>
> [Alia] Hmm, this ties into the discussion about how we want to handle
> redundancy and recovery for clients.   It's also a bit of a
tautology - a
> client is solely identified by its identity.    I have changed it to
say
> that "Each I2RS Client will have a unique identity" - but  that just
helps
> clarify the intent.

I think that this nicely encapsulates a confusion between identity and
identifier.  Identifiers identify.  Objects, in a very generic sense,
have identity.  Thus if a human being is an instance of an object, they
may be identified, based on context, by SSN, passport number, name, name
and date of birth, cell phone number etc; all could be valid
identifiers: but equally, a cell phone number could be the identifier of
a cell phone, which is associated with a function and multiple people,
while the cell phone could also be identified by its IMEI so the
determination of what is an identity, may take some consideration.  This
is often critical in security; you have a secure channel but with what?
Is the identifier sufficient proof of the identity?

Working with routers, you usually have multiple identifiers; the SNMP
sysName is not (usually) the OSPF 32 bit router id, while the BGP
Identifier (note, identifier) is different again.

Identifiers exist within a namespace, with rules about syntax,
uniqueness and so on (even if this are not made explicit).

The revised I-D contains
" A secondary  identity is merely a unique, opaque identifier ..."
and
"An I2RS Client may supply a secondary opaque  identity .. "

I think that most uses of the word "identity" in this I-D are actually
referring to "identifier" but at the same time, given that almost all
routers have multiple identifiers (as above), then this issue, of the
difference between identity and identifier needs making explicit in this
I-D.

Tom Petch

(p.s. if you have multiple virtual routers in one physical router, how
many identities are there? Discuss.)