Re: [sig-policy]Proposal prop-014-v001.txt - Lowering minimumallocation
> The reasons for suggesting a lowering of the minimum
> allocation are twofold:
>
> - in lowering the bar, more entities will qualify for
> address space. thus lowering the initial size was felt to
> be the appropriate thing to do in terms of responsible
> management and conservation.
i think that i can see the social good, and do not necessarily
disagree. but it seems to me that we are proposing to trade
conservation off against it, i.e., be less, not more
conservative, to achieve the perceived social good.
> - across the APNIC membership as a whole, the statistics in
> the paper show that many do not use up their first allocation
> within the one year time-frame and 63% only hold one
> allocation. So it seems that for a significant number of
> APNIC members, a /20 is more than enough as an initial
> allocation size.
ahhh. so this is what you mean by conservation, possibly
conserving address space allocated while possibly spending
routing table size.
> Examining the ASN allocation statistics for this region shows
> that the number of ASNs allocated is *way* below that of the
> other regions (with the exception of LACNIC).
yep. slides shown yesterday at ripe show apnic allocating more
space than any other region, but apnic allocating less
autonomous system numbers.
> Thus the degree of internetworking/ peering in economies
> across this region is quite low.
actually, we discussed this some yesterday here at the ripe
meeting. this conclusion is not clear. perhaps in the apnic
region there are less autonomous system numbers used per isp
than there are in the ripe or arin region? perhaps there is a
slower growth in the number of new isps?
we concluded that there needs to be considerably more study of
the data to understand what is actually happening.
> If allocations are restricted to those that are multi-homed
> the effect may be to exclude many more ISPs.
or may not be. how can we can really know? here in amsterdam,
it was felt that the bottom line is that we are all operating
on guesses from what may now be old and outdated assumptions,
and a fresh look at allcoation, conservation of space versus
routing table size, what enables new isps and services, etc.
> The question of 'what type of organisation can qualify' has
> always been left out, because in many cases it is very hard
> to define exactly.
>
> So yes, a side effect may be an increase in the number of
> organisations applying for allocations of portable space.
yep. but in that case, would the region not be seeing more
autonomous system number allocations?
> Part of the proposal refers to the results of the last change
> in the minimum allocation size. If you look at the data
> collected from the routing tables, the number of /19's routed
> is similar to the number of /20's routed. You might have
> expected to see many more /20's - but it seems that the
> policy change has not significantly added to the number of
> routes.
what is the distribution in age? in age of isp? e.g., what if
two years of /20s are the same number of chunks as ten years
of /19s (the danvers /19 agreement was in about '95 i think)?
note that i am not really for or against the proposal in
substance. i don't really know enough. i wonder if others
really do. are we shooting too much in the dark?
randy