ripe-589 question
Dear all, We've ran into internal conflict assigning IPv6 address space. Anyone caring to provide independent view on this would be much appreciated. Problem: LITNET (org: ORG-LA11-RIPE) has been comprised of two ASNs 2847 and 5479. Current IPv6 allocation is /29 (inet6num: 2001:778::/29). Current policy is to assign /32 per AS. Proposed new policy is to assign /30 per AS. Arguments for new policy: - RIPE-589 3.4 (Aggregation) and 3.8 (Conflict of goals) Arguments for current policy: - Two routes will be announced anyway. Different AS_PATH and routing policies, no aggregation. - /32 is 1200+ /64s per head of population of the country. Should be enough for any local AS for foreseeable future. Revise assignment planning if not. - The same address space allocation will be preserved for future AS if any current end-user (university) would require independent routing policies. - We will not get any wider allocation. Next address range (if it will happen) will be far off from 2001:778 I'm not in favor of wasting a long-term resource like IPv6 and rather deviate from the policy, but maybe I'm missing a point here somewhere ? TIA sincerely, Raimundas Tuminauskas KTU ITD / LITNET NOC Studentu 48a, Kaunas 51367, Lithuania phone: +370 37300033 fax: +370 37300643 email: raimis@litnet.lt
Hi,
We've ran into internal conflict assigning IPv6 address space. Anyone caring to provide independent view on this would be much appreciated.
Problem: LITNET (org: ORG-LA11-RIPE) has been comprised of two ASNs 2847 and 5479. Current IPv6 allocation is /29 (inet6num: 2001:778::/29). Current policy is to assign /32 per AS. Proposed new policy is to assign /30 per AS.
Arguments for new policy: - RIPE-589 3.4 (Aggregation) and 3.8 (Conflict of goals)
Arguments for current policy: - Two routes will be announced anyway. Different AS_PATH and routing policies, no aggregation. - /32 is 1200+ /64s per head of population of the country. Should be enough for any local AS for foreseeable future. Revise assignment planning if not.
I wouldn't count separate /64s as that will give you a distorted number. Count using the assignment size you are using, which I assume is a /48 per customer/university/research-institution/etc. That gives you a maximum of 65536 per /32. In a country with a population of 3 million that is probably enough to number all your customers :-)
- The same address space allocation will be preserved for future AS if any current end-user (university) would require independent routing policies. - We will not get any wider allocation. Next address range (if it will happen) will be far off from 2001:778
So why don't you announce 2001:778::/32 from one AS and 2001:77c::/32 from the other? If you need more space in one AS then you can grow to a /31 or /30, and if you don't need to grow them then you might, if at some point in the future you need a separate routing policy, announce the remaining space from a separate AS.
I'm not in favor of wasting a long-term resource like IPv6 and rather deviate from the policy, but maybe I'm missing a point here somewhere ?
Well, you can get the /29, and nobody else is going to get it, so you might as well make the best use of it. I just asked the NCC to expand my /32 to a /29. I use the first /32 for a LISP-based ISP setup, and I'm going to use one or more separate /32s for training purposes for ISPs. The nice thing about IPv6 is that we can always get enough space for what we need (within limits of course ;-) Cheers, Sander
participants (2)
-
Raimundas Tuminauskas
-
Sander Steffann