Hello ipsec-me,

I wanted to chip in as i am following up the debates on this list on
the direction the working group is going for. My personal interest
comes from the vendor side (ipsec/ike implementation in previous jobs)
as well as deployment and design (user side). Full disclosure: i work
for Cisco but very far away from product development/standardization,
i therefore speak for myself.

I think draft-detienne-dmvpn-00 should be the basis for
standardization work. My main arguments are:

-> clear layer separation: this project in fact create a layer 2
resolution protocol for NBMA networks, i don't see why this should sit
within a protocols that is made for key distribution and peer
authentication (and possibly security policy distribution, see next
point).

-> policy enforcement and distribution should not be a main driver for
ipsec/ike: today's requirements for security requires deep inspections
and all sort of other services (policy mgmt, enforcement and
distribution) to be ran across a vpn tunnel, we should focus on just
that: the tunnels.

-> the more the solution will be self contained in ike the more
proprietary extensions will show in the field so that vendors can add
their own features, this could affect interoperability.

-> topology discovery and reachability management are 2 different
things and i believe this proposal's goal is focused on dynamic
topology discovery and establishment. Routing protocols solves
reachability management if required (redundancy, fast convergence).

-> it can be deployed with today's ike / ipsec implementations (for
expl no kernel change in linux).

-> it does not require specific support for any non ipv4|6 unicast
protocols that could be ran on top of the tunnels (mpls / network
virtualization, mcast,  etc..).

-> sitting outside of ike means it can ran without it over ipsec with
manual keying. I know a lot of people disregard this but this means
this could be applied to other type of environment (military ipsec for
expl) which relies on physical keys.

-> Scalability / multi-hubs are taken into account.

thanks,

regards

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

Reply via email to