I had a conversation with a couple of folks while at IETF that they thought I 
should bring to the list for discussion.

While we have an operational considerations document that covers origin 
validation, it focuses mainly on policy and implementation
details of the validation machinery. We don't have anything that covers the 
back-end of implementing a proper RPKI (from the cache
upward, rather than downward towards the router). 

Put another way, a lot of the folks involved in this protocol's design and 
implementation are already intimately familiar with how
to manage things like ROAs to ensure that the underlying trust of the system is 
not compromised. 
I (and I assume other operators and likely the RIR staff) on the other hand. 
not so much. Keep in mind that while we might consult
with our security folks on implementation details, that doesn't necessarily 
mean that we'll do it right, since the level of
knowledge of the considerations here by the security folks may be limited. 
Since ultimately origin validation's trust level is only
as good as the trust level associated with management of the validation keys, 
I'm quite certain that additional rigor is necessary
in both the way that RIRs manage their POC records and authority to request, 
invalidate, and reissue keys, as well as how ISPs do
the same for their downstream customers, especially since these are likely to 
be managed through some form of automated back-office
system or customer portal.

So I think that there is a need for a document that covers things like identity 
and authority management, minimum levels of security
for key management, etc.

Basically, if you had to explain to someone how to ensure that the ROAs that 
are being delegated are:
1)      Delegated to the right person (right company, and person authorized to 
make the changes within the company)
2)      Have a reasonable assumption that they will not be compromised once 
delegated (proper key management)
3)      Methods for managing breaches, either due to failures in #1 
(disgruntled/former employee) or failures in #2
How would you do it? Are there existing documents we can use as a baseline and 
just fill in some of the specific details in the form
of a few different use cases, such as RIR to primary resource holder, resource 
holder to delegate, etc? 
An example to illustrate the reason for this - we have a subsidiary that we 
bought that had their own address space. Somewhere along
the way, a former employee changed the whois and POC info in ARIN's database to 
point to a new company, and said new company started
threatening to announce "their" address space that we were squatting on. 
Obviously a quick contact to ARIN sorted this out, but
under less amiable circumstances, they could have invalidated our existing ROA, 
issued their own, and created a serious problem for
us. While the solution is primarily that companies need to stay on top of their 
POC records now, we need to make that clear as a
part of the documentation that we're providing on how to implement this system 
and we need to try to give ideas on how best to
improve the level of trust.

I'm happy to help play the part of RPKI n00b to ensure that a draft written to 
cover this answers the right questions, but for the
same reason, I cannot help much with actually writing it, and am hoping that 
there are folks interested in picking up this work.

Thanks, 
Wes George 

Attachment: smime.p7s
Description: S/MIME cryptographic signature

_______________________________________________
sidr mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/sidr

Reply via email to