Re: [sig-policy] prop-053-v001: Changing minimum IPv4 allocation size to
Hi Toshi,
Interesting proposal. Comments in-line...
Toshiyuki Hosaka posted the following proposal on 8/1/08 17:07:
________________________________________________________________________
prop-053-v001: Changing minimum IPv4 allocation size to /24
________________________________________________________________________
1. Introduction
----------------
This is a proposal to change the minimum IPv4 allocation size from /21
to /24 and to create a new membership tier with an annual fee of
AU$500 for members with a /24 allocation.
This is cheaper than the existing lowest APNIC tier though? Which
doesn't make a lot of sense.
I would like to propose that the author simply proposes changing the
existing lowest APNIC membership tier to get a /24 (rather than no
resource at the moment).
2. Summary of current problem
------------------------------
In India, there are a lot of smaller ISPs who do not actually require
a /21. These ISPs would be satisfied with even a /24
I've yet to see an ISP who would be satisfied with a /24. NAT is not a
replacement for real address space. India has a population similar to
China, claims similar growth to China, yet can't even muster a single /8
from combining all the address space in use in the sub-continent.
3. Situation in other RIRs
----------------------------
The minimum IPv4 allocation sizes in other RIR regions are:
- AfriNIC: /22
- ARIN: /22 for multihoming, otherwise /20
- LACNIC: /20
- RIPE: /21
This info really doesn't help the author's case, does it. And I'd like
to assure the author that many small ISPs in Africa are a *lot* smaller
and a *lot* more needy than any ISP I've come across in India.
Out of curiosity, and hopefully someone from APNIC can help here, what
are the distributions of allocations per prefix size within the APNIC
region? (i.e. how many allocations are there at each prefix level)
4. Details of the proposal
----------------------------
It is proposed that:
1. The minimum IPv4 allocation size be changed from /21 to /24.
2. A new membership tier be introduced for /24 allocations.
This new tier will have an annual fee of AU$500.
See my proposed amendment above. It is much simpler.
5. Advantages and disadvantages of the proposal
-------------------------------------------------
Advantages:
- Small ISPs will be able to request an allocation smaller than a /21.
If small ISPs threw out their NATs, they'd be able to justify a /21
allocation very easily.
- More small ISPs will be able to afford direct allocations from
APNIC.
- IP resources can be saved by reducing potential waste associated
with giving a /21 to small ISPs that do not need that much space.
Why are we worried about saving IP resources when APNIC (and the other
RIRs) have a huge amount of IPv6 address space just waiting to be
distributed? ;-)
Disadvantage:
- No disadvantage to anybody.
This is a joke, isn't it? Either that or it displays a stunning naivety
of the Internet Routing system as it stands today.
There are numerous disadvantages:
- Internet Routing table bloat gets even larger (which it undoubtedly
will do as the market for IPv4 address space comes into being around the
time the RIRs have no more IPv4 resources to distribute).
- ISPs will have to spend more money with their favourite router vendors
if they want to multihome or participate in the default free zone (see
below).
- the quantity and frequency of BGP updates undoubtedly will increase
faster than they currently are increasing, as more and more smaller ISPs
contribute more and more smaller prefixes to the Internet routing
system. ISPs will have to buy bigger route processors sooner than they
expected.
- the greater number of prefixes means that some of the global carriers
may start filtering these small allocations, simply to protect their
routers and backbone integrity. So having an allocation will mean very
little as it won't be routable beyond the network neighbourhood.
There is another advantage:
- the router vendors make more money selling unplanned router upgrades
to ISPs around the world. Speaking briefly as an employee of one vendor,
this makes me happy.
In summary, while the proposal may be considered to solve a problem in
India, it has dire implications for the rest of the Internet. If it
results in India's /24 ISP members being filtered by the rest of the
world, what exactly will we have solved here?
6. Effect on APNIC members
----------------------------
A lot of new smaller members will join APNIC.
Why wasn't this listed as an advantage?
7. Effect on NIRs
-------------------
No effect.
It will have an effect. APNIC will then have a /24 minimum allocation,
which the NIRs will then have to consider implementing for their
membership too.
Hopefully the author will consider revising the proposal with the
suggestions above.
Best wishes!
philip
--