2 Aug
2021
2 Aug
'21
6:49 p.m.
On 3/8/21 6:29 am, Rosy Wolfe via 44Net wrote:
Hello everyone,
I, along with the board and staff, have been reading these messages.
First of all, I want you all to know that YOU ARE HEARD. The point of
having the TAC put out a proposal was to get feedback before adoption.
It turns out that a significant part of the feedback is negative. I
think that this proposal needs more work and adjustment before we can
consider implementing it. The board and I want to see consensus on the
main points of a proposal among the major schools of thought on this
mailing list. That said, it’s important to remember that the people on
this list are not the only people using the AMRPnet. We have a complex
task on our hands to reach as many of those people as possible as we
evolve proposals toward consensus.
I'm in two minds about the proposal, and it
comes down to "more
information needed" (more below). I'm one for whom renumbering will be
a major exercise, with 200+ IP addresses, plus having to liaise with
various network administrators for manual intervention and running some
geolocation sensitive services, which means a protracted changeover.
I'm not opposed to doing this, BUT it has to be worth my while, with
clear benefits for the network (and hopefully myself) at the end of it.
Last thing I want is to spend that time and effort, and be where we are
now, or worse, having to do it again, to fix up something unforeseen.
Several board members have suggested that it's hard to find consensus
on solutions until we have a consensus on what problem(s) the
solutions are trying to solve. We have a tangle of issues like the
complexity of IPIP tunnels, to BGP routing, to address space
sparseness, to low performance.
Yes, I'd like to see the problems fleshed out
first, so we have a clear
definition of the problems, their priority and proposed solutions.
With this in mind, what problems with the AMPRnet do you think we
should be trying to solve first?
One thing we haven't communicated well before, is that we are actively
discussing budget and infrastructure for a “backbone” network of PoPs
(Points of Presence) of the 44net on various continents, to make it
easier for hams to connect to the AMPRnet with minimal effort and
higher performance. If you have ideas about how you would like to see
this happen, feel free to share here on the mailing list. I know that
there’s at least one alternative proposal on the way.
First thing for me is to
replace the IPIP mesh with the proposed
backbone. While I have IPIP working here, I'm not 100% convinced that
it's as reliable as I'd like. Often, my routing table has only had
routes in 44.128/10, on occasions I've checked. While direct point to
point links should offer good performance, I'm not necessarily convinced
I'm seeing that, and there's many possible places things could break
(routers not passing the IPIP protocol, for example). The backbone/POP
idea could offer simpler setup for the cost of slightly less optimal
routing for end users. The VPN software can generally route subnets, if
told to do so by the server (perfect for those of us with subnet
allocations). The question is whether we can make it as easy for ARDC
to manage as the current system.
Other issues I see are:
Manual DNS management seems so 1990s. I haven't made any changes to my
DNS, as manual changes by a third party (network coordinator in this
case) is a major barrier. And a simple means of reverse DNS delegation
would be nice for those of us with /24 or larger allocations. And
having control of firewall separated from DNS would be nice - I may want
ampr.org DNS "internally", but not want my RF sites Internet connected.
There seems to be questions of how to communicate between intranet
(those on the radio/tunneled network) and BGP announced subnets, and
this is where different people may need different solutions, depending
on their network topology, etc. I have a private VPN between my
tunneled and BGP subnets, because some routes would otherwise go via San
Diego otherwise - that's working "long path" (~18000-20000km) for a
server 150km away! :) I'd like relatively right connectivity between my
BGP and intranet subnets, and possibly other BGP routed subnets, but no
connection (generally) to the wider Internet from my part of the intranet.
Anyway, just putting some thoughts out there for discussion.
--
73 de Tony VK3JED/VK3IRL
http://vkradio.com