[address-policy-wg] 2014-01 New Policy Proposal (Abandoning the Minimum Allocation Size for IPv4)

* Rob Evans

>>     http://www.ripe.net/ripe/policies/proposals/2014-01
> 
> Overall I think this is a good thing,

+1

> but I wonder if there is a reason for leaving 5.4 (minimum
> sub-allocation size) as-is?
> 
> If we open the door to transfer prefixes smaller than a /24, should 
> sub-allocation of them be prevented?

I think not, that would not be consistent. Maybe it's just an oversight
by the proposal's author to not have removed that particular paragraph?

> The routing side of me, of course, might consider the alternative of
> clamping the transfers at /24 too, but perhaps that should just be left
> for consenting adults to negotiate between themselves.

I say the latter. The current de-facto routing limit of /24 isn't
something borne out of our address policies, it's a limit the routing
community has come up with all on its own. So if we're putting in /24 in
the policy we're not really «clamping» anything, rather just documenting
today's operational reality (which could possibly change, in turn making
policy outdated).

Besides, it's only ALLOCATED PA and SUB-ALLOCATED PA inetnums that are
have a minimum size of >= /24 anyway. inetnums with status ASSIGNED PI¹
can be /29, ASSIGNED PA², as well as route objects³ can be any length up
to and including /32. Sky isn't falling yet though. :-)

[1] http://www.ripe.net/ripe/docs/ripe-555#longest-prefix-tables
[2] whois -T inetnum -x 185.47.43.255/32
[3] whois -T route -x 185.47.43.255/32

In the end I think the operators is perfectly capable of figuring out
what's the minimum size route advertisement they're willing to accept
without RIPE address policy instructing them (or pretending to, anyway).

Perhaps the ability to recycle infrastructure PI assignments as PA could
even be considered a small incentive for some operators to migrate some
of their infrastructure to IPv6, thus freeing up precious IPv4 resources
that can be then assigned to revenue-generating customers - alongside an
IPv6 assignment of course... :-)

Tore