Well I support the proposal El 14/02/2007, a las 18:41, Wilfried Woeber, UniVie/ACOnet escribió:
- "...when routing is a major issue for the End User"
there is no definition of "major issue" and there is no indication regarding which entity has to qualify or check this statement.
My understand of "major issue" is "when the end user is going to use it for routing to/from Internet"
In effect this will open a path to a minimum /24 assignment per site, and PI instead of PA.
Not a bad idea. If a end user wants multihoming, he will get it (becoming a LIR itself, getting a slice of a PA and announcing it independently,etc.: money talks and RIR listen) and probably (I think so) this is the least dangerous way to do it. It will increase the routing table size like any other but at least this will help to preserve some address space.
- when the ISPs decide to NOT route some small address blocks then trying to circumvent their configuration and intent by messing around with address assignment policies is royally broken. If this is really an operational problem then it needs a resolution on the routing plane.
I dont see it as an operational problem. I see it as a business problem. Companies want/need multihoming and will get it one way of another. We can set up things to do it in the least dangerous way or try to oppose it and see how things get worse.
- on a more general note, as long as the minimum assignemt size for customers receiving PA is raised to /24, too, this proposal is a real incentive to go for PI instead of PA.
I agree with that. I don't see a real reason to assign lower than /24 to any customer (and assigning less than /24 is a nightmare for reverse DNS). My two cents.
------------------------------------------------ Tecnocom Fernando Garcia Eurocomercial - Depto Técnico Josefa Valcárcel 26, Edificio Merrimack III Madrid 28027 Tel: +34 91 4359687 Fax: +34 91 4313240 e-mail: fgarcia@eurocomercial.es http://www.tecnocom.biz http://www.eurocomercial.es