Search RPD Archives
[rpd] Some thoughts, and some actions required
owen at delong.com
Wed Feb 3 19:23:06 UTC 2016
> On Feb 3, 2016, at 06:06 , Andrew Alston <Andrew.Alston at liquidtelecom.com> wrote:
> You say that but history also indicates that there have been significant issues in certain cases around universities applying for large tracts of space. Most of which were resolved at the time, but it was a fairly complex exercise that at times took long periods of time and a lot of back and forth that I’m not sure most universities are geared up for.
> Sadly, our current policies are such that there is enough ambiguity that large applications often require explanations which can at times be fairly difficult to calculate and provide, for example, the issue of wifi usage concurrency has proved to be a major point of contention, because on a new wifi deployment its extremely difficult to prove concurrent utilisation without historic figures.
While I did not support the educational policy because I did not believe it appropriate to give special privileges to educational institutions to the detriment of other classes of community members, I would support a policy that generally allowed for a reasonable estimation of concurrent wifi users based on quantified total user base.
For example, if reasonable evidence could be presented for a probable userbase of 50,000 wireless devices, I would support a policy which allowed for an assumption of up to 80% concurrency (e.g. 40,000 addresses needed) with an appropriate round-up to bit boundary for IPv4. (In the case of IPv6, we’re counting subnets, not
devices anyway, so I don’t think it matters).
> The other issue in the past (which thankfully seems to have been resolved) is the debate about if a university is an EU or an LIR, I remember debating this at length with AfriNIC a few years ago and there was substantial discussion on this list around this point (please lets not reopen this one, it gets messy)
It gets messy because most universities want to act as LIRs while being treated as EUs. It gets messy because there has never been a good clear definition of what constitutes an EU vs. an LIR and many seek to muddy the waters based on the reporting of reassignments/reallocations rather than looking at the direct fact of whether addresses are reassigned/reallocated. It gets further muddied by those who would seek to claim that dynamic reassignments/reallocations do not constitute reassignments/reallocations.
It would be very clean if we defined it as follows:
An LIR is an organization which provides addresses (regardless of mechanism) to third parties for use on equipment neither owned, nor controlled by the LIR.
An EU is an organization which does not in any way fit the definition of LIR above.
However, since most Universities are horrified at the thought of being considered LIRs, these definitions receive significant opposition.
> I to would support large amounts of space being left in the hands of academia, and it really saddens me that I know of universities that are still using NAT to push entire campuses through address space that is smaller than a /28!
Yes… This should be fixed to the greatest extent possible while there is still some extent to which it can be fixed.
> Why didn’t the policy pass? That I have never understood, because I never truly understood the arguments against it (for example, one argument that was put forward was that increasing the number of live IP addresses available to universities would increase spam on the internet)
There were a number of nonsensical arguments in opposition. My primary reason was that the policy wasn’t generally fair across the board. It granted an unfair advantage in streamlining the process for academic institutions to the potential detriment of other classes of community members seeking addresses.
> But, it is what it is and the community could not reach consensus, so we have to move forward and we’re moving into an arena of unchartered territory. I’m very curious to see what happens the day an African ISP applies for space and gets told there is no more, and the reason they cannot have any is not because their application is invalid, but because there is no more. I’m even more curious to see the reactions if there is no transfer policy and as a result zero ability to get their own space. I pray I never see my curiosity satisfied on the latter, and realise that its only a matter of time before my curiosity is satisfied on the former.
I think what will happen on the former is relatively predictable. I think it will largely be identical to what has happened in the other four regions as that event has occurred:
+ An uptick in the rate of IPv6 adoption
+ A mad scramble by the provider to clean up their existing space and optimize its utilization
+ The unfortunate deployment of CGN “solutions”
It’s true that the latter case will be more interesting and is the only actual uncharted territory in question. That will truly be interesting if it happens. It could well be that AfriNIC gets usurped by some private organization willing to register transfers. It could be that several such organizations spring up and the uniqueness of IPv4 address registrations within the continent becomes quite a bit more difficult (this would most certainly be a “bad thing” (tm)). I’m genuinely uncertain what happens without a transfer policy at that point. I’m also certain that life will be less exciting for those who have deployed IPv6 than for those who have not.
More information about the RPD