[sig-policy] Proposals prop-063 and prop-070

  • To: sig-policy at apnic dot net
  • Subject: [sig-policy] Proposals prop-063 and prop-070
  • From: David Woodgate <David.Woodgate at telstra dot net>
  • Date: Tue, 24 Feb 2009 11:38:27 +1100
  • Delivered-to: sig-policy at mailman dot apnic dot net
  • List-archive: <http://mailman.apnic.net/mailing-lists/sig-policy>
  • List-help: <mailto:sig-policy-request@lists.apnic.net?subject=help>
  • List-id: APNIC SIG on resource management policy <sig-policy.lists.apnic.net>
  • List-post: <mailto:sig-policy@lists.apnic.net>
  • List-subscribe: <http://mailman.apnic.net/mailman/listinfo/sig-policy>, <mailto:sig-policy-request@lists.apnic.net?subject=subscribe>
  • List-unsubscribe: <http://mailman.apnic.net/mailman/listinfo/sig-policy>, <mailto:sig-policy-request@lists.apnic.net?subject=unsubscribe>
      Proposals prop-063 and prop-070 both suggest increasing the restrictions on allocations of IPv4 addresses as IPv4 exhaustion approaches. prop-063 changes time restrictions of allocations, and prop-70 changes the size
      The key question for both proposals is: Does the exhaustion of IPv4 address pools warrant such increased restrictions on allocation criteria? (I.e. How do these proposals actually help the industry situation?)
      I believe the answer is that an increase of restrictions is only valid if it is likely to result in increased efficiency in the actual use of the remaining address space. If any proposed changes do not achieve such increased efficiency, then adopting them would simply increase the industry workload cost (both of the APNIC secretariat and the requesting ISPs) without any resulting industry benefit.
      Both proposals seem to me to imply that forcing a greater frequency of requests will result in a more efficient allocation of addresses. I broadly support this view, up to a point - actual address demand could change significantly over 12 months in comparison with the initial forecasts when requests were made, and reducing the maximum size of allocations might reduce any quantisation factors like any possible tendencies (if there are any) to round up requests to bit boundaries.
      However, any changes increasing the number of requests or limiting allocation sizes needs to: (a) Be able to be accommodated within the workload achievable by current APNIC staffing levels; (b) Allow the address allocation system to continue to function in a practical way for all of its members. - (I.e. Allocations must remain of a size and duration that have practical usability)
      Regarding prop-063, I believe that reducing the maximum duration to six months is a reasonable approach, but it needs to be demonstrated that any potential increase in the number of requests could be managed by the current APNIC hostmaster staff. (Hopefully, this wouldn't simply double the number of requests to be processed in a year?) This requires an estimate to be made of this likely increase; could such an estimate be provided by someone from APNIC staff?
      I likewise feel that the concepts suggested in prop-070 should be considered, but I believe that more work needs to be done on estimating the impacts of such limits to both APNIC and member workload.
      For example, if a maximum per request limit of /16 were introduced, then a member who needed a /13 would need to apply 8 times during the year - and while that may or may not be reasonable for the member themselves, if 100 members needed a /13 for the year then that would increase requests from those members from 100 to 800, which would be a significant workload increase for APNIC.
      Similarly, if the maximum size were /18, then a member needing a total of /12's worth of addresses in a year would need to submit 64 requests during that year to satisfy that demand, or significantly more than 1 request per week.
      I think that a distribution showing (Size of request vs. number of 2008 requests of each size) would be useful in terms of identifying (anonymously) both the number of members that could be impacted by this proposal, and also the potential overall annual increase to APNIC requests. This data might allow the tailoring of the scale described in prop-070 to optimise the benefit of the proposal against the cost of the additional work involved. Again, could relevant APNIC staff please supply such information to the list?
      In summary, I believe that both prop-063 and prop-070 have potential merit, but both proposals need to be assessed for probable impacts on APNIC and member workloads, and the detail of the proposals may then need to be adjusted to take such impacts into account. I therefore request the APNIC secretariat to provide to this list: (a) An assessment of the likely increase in number of addressing requests arising from prop-063; (b) The (anonymous) data of the number of requests in 2008 identified by size of requested address block; (c) Any assessments as to whether APNIC staffing would need to be increased to manage the likely workload changes should either prop-063 or prop-070 (or both) be implemented as written.
      			David Woodgate