[ppml] Proposed Policy: IPv4 Micro-allocations for anycast services
Alec H. Peterson
ahp at hilander.com
Fri Aug 12 11:47:59 EDT 2005
--On August 12, 2005 8:40:43 -0700 David Ulevitch <davidu at everydns.net>
> As do I -- and that's the problem. I worry about all of ARIN
> membership all of the sudden realizing that they too have a "real
> need" for anycast services. I think people have this impression that
> anycast == higher uptime. If anything that is probably not the case
> as it can easily be executed/deployed improperly and/or poorly.
That's like saying that the impression that multihoming == higher uptime is
a fallacy because it can easily be deployed poorly. Anycast does increase
uptime when done intelligently.
However, that notwithstanding, I think the concerns about potential abuse
of this policy are justified, but instead of claiming that anycast is too
hard we really should try to craft the policy such that abuse is not
For example, the issue has less to do with an entity not being able to
justify enough space for their entire enterprise and more to do with the
fact that they cannot carve a /24 out of their space for fear of having it
filtered. What if an org was required to already have an ARIN allocation
under the normal criteria to qualify for an anycast block?
More information about the ARIN-PPML