Re: doi-07/interoperability questions

[Yan-Fa LI <[e-mail]>]

If this assertion is correct, as a customer not a developer,
I see a use for option 4 and would ask you not to remove it.
Crypto arguments aside, there are three uses for this within 
the organization I work for.

One, is the obvious situation where I have no other option, because of
regulatory reasons, but I want improved IP where privacy is less
important than integrity.  For example, managing someone elses network
across a less trusted network, note this managed network could even have
non-publically routable address space.  This would be end-to-edge.

Two, improving leased line security from business partners, in an
end-to-edge situation.  Privacy is outweighed by authentication and
integrity requirements with less client and gateway processing.
This would be aided and abetted by simple firewall processing at
the end of the tunnel on the edge hardware.

Three, some xDSL providers are offering backhaul connections to
corporate networks.  Supposedly the traffic is across ATM VCs and so is
as private as leased lines are today, questionable security premise I
know.  If we decided that privacy were not an issue, then it would fall
back to authentication and authorization issues.  AH, in tunnel mode,
would suffice for this purpose, without the overhead for encryption;
would this be worse or better from a client standpoint than ESP 56bit w/
HMAC and replay with SHA-1 ?  I'd love to hear more expert opinion about
this issue.

Yes this is all end-to-edge centric.  I'm thinking clients to gateways
not gateway to gateway as yet.  That's phase II ;)

my $0.02

Y


> Sounds to me you are suggesting the following changes to the arch spec
> in section 4.5 Case 1. 
> ] 
> ]                   Transport                  Tunnel
> ]              -----------------          ---------------------
> ]              1. [IP1][AH][upper]        4. [IP2][AH][IP1][upper]
> ]              2. [IP1][ESP][upper]       5. [IP2][ESP][IP1][upper]
> ]              3. [IP1][AH][ESP][upper]
> ] 
> 
>                   Transport                     Tunnel
>              -----------------             ---------------------
>              1. [IP1][AH][upper]   (remove)4. [IP2][AH][IP1][upper]
>      (remove)2. [IP1][ESP][upper]          5. [IP2][ESP][IP1][upper]
>              3. [IP1][AH][ESP][upper] (add)6. [IP2][AH][ESP][IP1][upper]
> 
> Is this correct?
> 
> I think it is ok to remove 4, it really doesn't buy you much.  
> I think we should keep 2.  This new one for tunnel mode seem 
> to make sense.  Now, should we restrict 6 to just gateway-to-
> gateway?
> 
> /eric