Re: [stir] current draft charter

["Wendt, Chris" <[e-mail]>]

Thanks Brian. 

Answer inline.

On Jun 13, 2013, at 10:08 AM, Brian Rosen <[e-mail]> wrote:

> The IETF is starting a new effort on routing,  see the TERQ work.  If you have ideas on routing, that would be the place to bring them.
> 
> I think the problem we have here is trying to decide who is a "good guy".  Most of the service providers who would would call "pink" in this context would claim they are good guys.  Trying to create a mechanism to separate good guys from bad guys is very hard.
> 

Sure, they assert their domain, provide DNSSEC to verify, but we notice customer complaints and shut them down.  Plus have cryptographic evidence that it was indeed coming from them.  I think that works pretty well.  
If a "bad guy" chooses not to participate in asserting domain, then the service provider can choose whether to accept the calls.

> I'm not sure why protecting TNs individually is more complex then the view you have of routing, especially because we have tried what you are suggesting, although we did not have DNSSEC.  Dealing with the privacy issues, just to start, would be the first real barrier.

I probably don't fully understand the privacy issue, but to me the only thing potentially exposed is the domain that owns the telephone number.

> 
> Let's say we got beyond that.  We then need someone, probably someone like Neustar, to maintain a per country portion of the database.  How would it do that?
> 
> Basically, every time a number was delegated, it would change the domain of the entry to reflect the delegation.  That could work.  What it means is that delegation of any kind has to be recorded in the new database.  If you note, certs follow delegation, with you inserting "domain" in the middle (domain follows delegation).  The twist is that instead of a 2nd level reseller issuing a cert to a 3rd level reseller, the 2nd level reseller goes to the new database and has them effectively do the same thing.   

So, I defer here to better experts than myself.

> 
> Having databases that associate a domain with a TN might help things like finding a cert, but it doesn't affect the basic issue of how you avoid spoofing.  Do you think the approach we have suggested here (signing of From/To/Time/Session Id) with a cert traceable to the identity is appropriate?

I just think it's overboard to have a cert per identity and I don't believe a service provider I want to police calls TN by TN.  
It's the domain representing the wholesale provider or PBX or whatever that we want to police, I think.

> 
> If you are "simply" suggesting the cert is located via DANE in the domain of the caller, then for user@domain, I think we're probably saying the same thing.  Where we seem to differ is that you suggest we have a DNS entry for the TN that has the domain in it, and then you use that to find the cert used for signing.  In the basic case, we're suggesting a URI to the cert in the signaling, which seems simpler.  What matters however is how the DNS tree for the TNs is maintained.

I think using standard DNSSEC mechanisms is simpler and more straight forward, but that's an opinion guided by domain routing as the end state.
Although, I think perhaps it might be wise to have a guiding philosophy that I hope most would agree with.  If we are creating new mechanisms to do things, we are doing it wrong.

> 
> So, my response is the same as I have given before: the privacy and operational issues of putting per TN information in the public DNS are overwhelming.  We have tried it, and it has failed.  

I think the main issue, to your point above, is management of a TLD, but i think it's solvable, we have lots of smart people.  
If the privacy issue is really an issue, then lets discuss further.

This will be a transition from private databases to a fully distributed system, and I think the long term issue will be the desire to move to this model or a push from regulatory entities if that is where we/they want to go.  In some sense, I think there will be a lot of benefit to it and allow for much better communications applications in the future, but this is speculation.

In the end, we have a great distributed internet model for transporting data, securing data, cryptographic verification of destination and origination, why shouldn't we push that direction for communications as well, isn't that the point?  Or better yet, why won't it be the end state anyway whether the traditional players are involved or not.

Failure is not an option :)


> 
> Unlike Hadriel, you really do want to disclose domain with TN.  That has been a non-starter for many carriers.
> 
> Brian
> 
> 
> On Jun 13, 2013, at 9:37 AM, "Wendt, Chris" <[e-mail]> wrote:
> 
>> Hi Folks,
>> 
>> I believe it couldn't be a better time to start discussing public routing.  While this is a little beyond the scope of protecting caller-id, I think the effort is highly complimentary, and solving both together will provide a better path to get things implemented in service provider networks.
>> 
>> IP communications is evolving whether you want to get on the train or not.  I applaud Henning's efforts to push the ball forward.
>> 
>> Moving more to an internet model for routing is inevitable, IMHO.  There is questions about how relevant TN is or will be in the near future.  For the sake of having any chance of success, let's simplify the effort and focus on the future.  There will be a transition process to get there, no doubt, but let's focus on the end goal, and not focus on fixing problems of the past.  We are already in a world of multiple private peering arrangements, LCR, and managing multiple egress points anyway, so the transition should not be that difficult.
>> 
>> I believe a common routing mechanism for both user@domain and TN/e164 based identities should be the focus.  Treat both as first class citizens for IP communications.
>> 
>> Use DNS and domain based routing for all.  Use DNSEC to validate and cryptographically associate TN/caller-id with a domain (for good guys) and do domain routing from there.  
>> Forget about trying to protect TN's individually, that is too complex.
>> 
>> It's up to the consumer service providers to enforce and block bad things from bad domains.
>> It's up to wholesale/trunking providers to allow customers to assert their own domain, or be responsible for asserting domains on their behalf at the risk of being considered a "bad guy".
>> 
>> This is high level, and perhaps a sunny day view of the world, but I don't think unrealistic.  Frankly, if there is any chance of a common federated communications system continuing to be relevant, I believe a simple approach that matches the internet model is the only choice.
>> 
>> You should see more concrete opinions from Comcast in the future, but I wanted to take the opportunity to express these thoughts and would love feedback on why this should or should not be the model going forward.
>> 
>> If there is agreement, I believe some of this should be addressed in the charter, because I think some of the assumptions may not enable this path. 
>> 
>> -Chris
>> 
>> 
>> 
>> 
>> On Jun 13, 2013, at 8:51 AM, Dan York <[e-mail]> wrote:
>> 
>>> Hadriel,
>>> 
>>> On 6/12/13 5:38 PM, "Hadriel Kaplan" <[e-mail]> wrote:
>>> 
>>>> That begs the question of what issues you think made Public ENUM fail,
>>>> and why we won't hit the same issues in whatever model we choose.
>>> 
>>> Since you addressed this question to me, I feel compelled to answer... but
>>> in the meantime both Brian and Jon have given much more detailed answers
>>> with which I mostly agree.
>>> 
>>> I saw security/privacy issues as the main issues that made Public ENUM not
>>> work, i.e. Brian's points 4, 5 and 6:
>>> 
>>>> 4. Any public database that was able to show any information about a
>>>> telephone number was considered a privacy issue, requiring a lot of "sign
>>>> off", which never happened.
>>>> 5. Everyone objected to being able to determine what numbers were "live"
>>>> 6. Carriers objected to a public database that told competitors what
>>>> numbers they controlled
>>> 
>>> 
>>> The potential for spam was also high.  Five or six years ago there was at
>>> least one tool circulating around that would walk an ENUM tree, generate a
>>> list of all potential phone numbers and then send SIP INVITEs to all of
>>> those numbers.  I remember either seeing a video or reading about how
>>> someone did this with all the public ENUM numbers published for Germany
>>> (at that time). So using Public ENUM could be a fantastic way to
>>> potentially get yourself on the calling lists for telemarketers.
>>> 
>>> Ideally, I think a public service *would* be a useful way to enable
>>> ubiquitous usage. I'm just not sure how to get there without also opening
>>> a solution up to these same kind of security/privacy issues.
>>> 
>>> Dan
>>> 
>>> --
>>> Dan York
>>> Senior Content Strategist, Internet Society
>>> [e-mail] <mailto:[e-mail]>   [phone]
>>> Jabber: [e-mail] <mailto:[e-mail]>
>>> Skype: danyork   http://twitter.com/danyork
>>> 
>>> http://www.internetsociety.org/deploy360/
>>> 
>>> _______________________________________________
>>> stir mailing list
>>> [e-mail]
>>> https://www.ietf.org/mailman/listinfo/stir
>> 
>> _______________________________________________
>> stir mailing list
>> [e-mail]
>> https://www.ietf.org/mailman/listinfo/stir
>