Re: [apnic-talk] Questions on microallocation assignment policy
I think we need to be a bit clearer on what the effect of the /24 filters
would be.
Would the proposed filters cut these micro-allocation /24 off at the
knees from being globally routed? No. I don't think so. I think, they
would prevent somebody from making semi random assertions of
connectivity in respect of /24s from *bigger* allocatiuons, NOT from all
footprints handed out directly by the RIR.
If the RIR tell the wider community that a given range is allocated on
a boundary of /22 or /23, then that becomes the allocation boundary for
filter rules in respect of those addresses. We just need this documented
and some process for change.
If you get your hands on a /22 from somewhere else, and de-aggregate it
on the wider horizon of visibility, then bets are off.
I see this policy as another side of both NOPEER and filtering. Even though
there appears to be conflicts, they are not in contradiction:
+ don't de-aggregate your assignments on the wider horizon
o use NOPEER to de-aggregate them on peer horizon boundaries
to achieve localized visibility multihoming for resilience.
+ if you apply for micro-allocation, you can't expect to get
larger aggregatable space in the same block. since its a micro,
you would expect renumbering costs are constrained. So, if you
apply for this space and can grow beyond it, accept renumbering
as a neccessary evil for you, in this special case. But, you
get addresses, and you get them to scale your operation/budget.
+ the filter rules are not a rule applied by RIR, but a pragmatic
decision the wider routing community can make to limit their
costs of your routing strategies. If you want to change your
visibility under an RIR allocation horizon, pay peering costs
or customer costs, and use NOPEER.
is that the kind of policy outcome?
cheers
-George
* APNIC-TALK: General APNIC Discussion List *
* To unsubscribe: send "unsubscribe" to apnic-talk-request at apnic dot net *