[arin-ppml] 2008-3 Community WIreless Networks

George, Wes E [NTK] Wesley.E.George at sprint.com
Wed Aug 19 09:55:29 EDT 2009


Owen, thanks for the clarifications.

I was the one that suggested the /64. However, I didn't say that they should *only* be eligible for a /64, only that under the pretense that they were not globally routed, we should start with a /64, with justification allowing larger allocations as necessary. Needing to delegate blocks to users is a valid justification for needing more than a /64. If you agree that assuming these will never show up on the Internet is a stretch, then a /48 probably makes sense as the minimum. I'll dispense with rehashing the obvious concern about routing table growth for now.

However, this confusion illustrates my problem with this proposal - no one can seem to even define what is and is not a community network, what purpose they might serve, what addresses and scope they might or might not need, and how to make the definition narrow enough to prevent abuse, while still being wide enough to be useful to a critical mass of the intended group. Therefore no one can come up with a universal justification for why globally unique PI space (and/or a reduced fee) is a requirement for all (or even most) of these applications and more importantly, what makes them unique compared with other ARIN members who are expected to qualify under the existing rules.

Most of the examples you cite are community networks of the rural ISP variety. This means that in order to provide service, they upstream connectivity to the Internet. Whether they get it for free/cheap or pay full price, I have trouble believing that they can provide a usable service (even when viewed through the "you get what you pay for" filter) if they don't have a stable connection to one or more providers that would be willing to give them address space. If this is just about trying to prevent unnecessary renumbers, I'm failing to understand why this would be different from any other small ISP that doesn't qualify for ARIN PI, pays full price for their upstream, and has to renumber out of their current PA space if they change providers. We don't tend to make exceptions for non-profits in other sectors, so is this about ARIN trying to advance the cause of universal Internet service by making it easier for these would-be ISPs? Seems awfully far into the political realm for an org like ARIN for my taste.

I agree with Martin that we should probably abandon this proposal and try again.
Please don't misinterpret this as me being against any proposal of this type, as I can see genuine value in having address space available for the hobbyists/experimenters/innovators out there. As I said in my previous mail, I think that we'd be better off with a proposal that allows for direct PI allocations for experimental networks, and then make the requirements such that some subset of community networks, especially the ones that don't necessarily count "ISP" among their primary objectives, would qualify to get space that way.
It's semantics, but I think it addresses an important concern that keeps coming up in this debate - not being so stingy with the very large address space that is IPv6 that we stifle innovation, while not being wasteful or opening up a loophole to provide everyone with a router and a non-profit business plan a /48 to announce to the global Internet.
Perhaps an update to NRPM section 11 with a slightly broader definition of experimental? Specifically, I'm thinking about allowing for open-ended experimental plans, with some provision to revisit the progress and justification for the experiment periodically to see if the allocation is still justified.

If we are genuinely concerned about the cost of the ARIN fees for small/experimental/hobby/community networks, perhaps we also need a separate policy that covers some manner of need-based exceptions to allow for reduction in fees for worthwhile causes.

Thanks,
Wes George
-----Original Message-----
From: arin-ppml-bounces at arin.net [mailto:arin-ppml-bounces at arin.net] On Behalf Of Owen DeLong
Sent: Tuesday, August 18, 2009 7:00 PM
To: ARIN PPML
Subject: [arin-ppml] 2008-3 Community WIreless Networks

While I would agree that 100% volunteer is probably too far, I think
it's better
than not getting any policy.

The AC was not able to agree on any course of action for this policy
after the
last public policy meeting.

Personally, I feel that this policy is needed, and, that there are
good reasons
for community networks to get provider independent space.

Community networks provide a useful service to a variety of populations,
many of whom would not otherwise have access to the internet. They also
tend to be in a position to provide communications resources in times
when
more traditional means of communication may not be as readily available,
such as in times of disaster.

Many of these networks depend on the good will and support of local
or nearby providers and receive donated connectivity. In those cases,
having portable addresses and the ability to exchange routes with one
or more such providers and the ability to readily accept connections and
disconnections as they come is a very useful thing.

I think that claiming these networks would not be advertised on the
global
internet was a mistake.  However, there are cases where community
networks
need to interact with multiple organizations in a way that isn't
entirely compatible
with ULA while still not being globally advertised.

One person suggested that a /64 would be adequate for 100-200 users.
This
makes no allowance for the fact that each of those 100-200 users may,
themselves
need a subnet.  In IPv6, a subnet is supposed to be a /64, so, that's
at least a /56
in any case and still doesn't allow for the networks necessary to
number the
infrastructure of the community network itself.

Many of these networks are not a single router sitting in a closet
somewhere,
but, have significant metro-area infrastructure, often involving
multiple
wireless links, VPNs, tunnels, and other diverse topologies.

One example of a community network that might put this in some
perspective can be found here: http://www.svwux.org

There are many, many more examples available, and, they are very diverse
in their designs, users served, and organizational structures.  About
the only
thing most of them have in common is a need for portable stable
addressing.

I can understand resistance to this idea in the IPv4 world, but, in
the IPv6
world, it just doesn't make sense to prevent these organizations from
getting
the addressing they need.

Owen


This e-mail may contain Sprint Nextel Company proprietary information intended for the sole use of the recipient(s). Any use by others is prohibited. If you are not the intended recipient, please contact the sender and delete all copies of the message.




More information about the ARIN-PPML mailing list