Hi, On Thu, Jan 08, 2004 at 11:01:09AM +0000, Michael.Dillon@radianz.com wrote: [..]
When some part of the infrastructure of the Internet is considered to be critical, the people responsible for it will build it so that it can never fail. They do this by applying fault tolerance techniques. Anycasting DNS is a fault tolerance technique. DENIC is asking for a policy that understands the need for fault tolerance when an organization is managing part of the Internet's critical infrastructure.
If this policy focuses on the fault tolerance technique then it is doing the wrong things. Policies are not about technology. RIPE policies are political agreements that balance the needs of everyone in the RIPE community. This policy needs to focus on defining which kinds of "critical infrastructure" are important enough for the whole community to justify special allocations.
The policy needs to balance everyones needs (among that: "little extra routes in the DFZ"). The policy does also need very clear-cut criteria to *decide* whether something meets the policy or not. Applying technical criteria is easy (easier, at least) than a fuzzy term like "critical infrastructure" that mean something different to whoever reads it. Commenting on your first example: for the USA, something might very well be a "critical infrastructure", like the US power grid, and *still* the rest of the world might not care much if it breaks down... - so the definition of "criticial infrastructure" is very much localized and fuzzy. Gert Doering -- NetMaster -- Total number of prefixes smaller than registry allocations: 57882 (57753) SpaceNet AG Mail: netmaster@Space.Net Joseph-Dollinger-Bogen 14 Tel : +49-89-32356-0 80807 Muenchen Fax : +49-89-32356-299