Re: [pcp] [ Side-by-side or nested protocols

[Margaret Wasserman <[e-mail]>]

Hi Alper,

On Sep 18, 2012, at 2:58 AM, Alper Yegin wrote:
> Again, I'm confused. With PANA, you get re-use of spec, implementation, security analysis. 
> Which one of these do you get with yet-to-be specified EAP-over-PCP?

Just to clarify -- you only get "implementation re-use" if you are able to use an existing PANA implementation on your platform.  That would require that the existing implementations:

- exist
- integrate well with your PCP implementation
- support the features that you require

People should read the long message that Sam Hartman sent to the list about existing PANA implementations, and make their own determination about whether one of the existing PANA implementations would work for them.  Two issues that are important here, IMO, are: 

- whether PANA brings in more complexity than is needed in the PCP case, because it is designed for the (more complex) network access case.
- whether the limitations of the existing PANA implementations (i.e. limited EAP methods support) are acceptable for PCP

If you won't be able to reuse an existing PANA implementation, then the "implementation re-use" benefit doesn't exist.

I don't understand how we can re-use the security analysis, since the PCP case is very different from a network access case.  In fact, the security requirements for PCP are very unusual, and quite specific to PCP.

Code reuse is also very desirable.  In any of our proposed cases, we could get code-reuse for the EAP portion on systems that use 802.11 security mechanisms or use EAP for other reasons. It is potentially possible that you could get additional code-reuse in the PANA case, if systems end up using PANA for more than one purpose.  Right now, though, there isn't much overlap between the types of systems that use PANA and the types of systems that are likely to use PCP.

None of this reuse is valuable at all, though (again, IMO), if it presents the user with an unintegrated experience and/or results in a hard-to-support or hard-to-maintain product.  Running two different protocols side-by-side and trying to present them as a single function to end-users and administrators presents various system design challenges that need to be weighed against the complexity of performing in-band PCP key management.

Margaret