Re: [Ideas] Spencer Dawkins' Yes on charter-ietf-ideas-00-00: (with COMMENT)

[Tom Herbert <[e-mail]>]

On Wed, Sep 13, 2017 at 8:16 PM, Spencer Dawkins at IETF
<[e-mail]> wrote:
> So, responding to Uma's response to Tom's response to my response to the
> proposed charter (whew!),
>
> On Wed, Sep 13, 2017 at 3:56 PM, Uma Chunduri <[e-mail]>
> wrote:
>>
>>         > Is a look at general security implications, in a form that
>> specific
>>         >framework  usages can point to, on the table for IDEAS?
>>         >e
>>         Spencer,
>>
>>         I believe there are two discrete components being championed in
>> IDEAS:
>>         One, is mapping system of identifier to locators and the other is
>> introduction of identity mapping. The former looks much more like a routing
>> or name resolution protocol, and the later would be doing identity
>> management and possible collecting PII. There are obviously many security
>> implications to      both parts, however I think the threats and sensitivity
>> between these is quite different, i.e. hacking into the ID/loc mapping data
>> base could result in misdirecting packets, hacking into identity store may
>> result in loss of users' privacy.
>
>
> Tom's response to me makes sense.
>
>>
>> [Uma]: Tom, you summarized well. I would note there is interconnected
>> aspect to these 2 items w.r.t security. Identity AUTH can inherently bring
>> security (and if needed privacy) to Identifier/Location mapping and
>> strengthen that area tremondoesly.
>> However, Identity privacy itself has  to be tackled and there are existing
>> well defined mechanisms for that as discussed earlier in the IDEAS list
>> (pointer from Diego, is a great example).
>> When we described identity and it's uses here
>> https://tools.ietf.org/html/draft-ccm-ideas-identity-use-cases-01#section-7
>> , we noted threat analysis aspect in Section 7 and was reflected in charter
>> too.
>
>
> Uma's response to Tom makes sense.
>
>>
>>         These seem fundamentally different so security considerations
>> should probably be considered independently of each other.
>>
>> [Uma]: Different but interdependent on some aspects as mentioned above.
>
>
> So, what I'm not understanding, is that there are two work items, and only
> one framework deliverable. Is the intention that the identifier/locator
> mapping system and the identity mapping system are different enough to have
> different security considerations, but are so tightly interwoven that
> neither is usable without the other, or with any other mapping system
> separately, so it makes sense to lump them into one framework?
>
Spencer,

The underlying mapping system (basically specialized key/value store
with protocol) is common between the two use cases. However, the these
cases are independent in a least the sense that identifier/locator
protocols (as well as those for network virtualization) require
identifier->locator mapping but don't require any concept of identity.
I believe, the need for the ID/loc mapping is well understood at this
point, whereas the concept and rationalization for identity in this
area are still emerging. Both of these are valid items to pursue, but
I hope that a WG can minimize dependencies so that development of the
ID/loc mapping system does not become a long, drawn out multi-year
process. If this can be done in a single framework then that would be
okay I think.

Tom