Re: Consensus Call: draft-weil-shared-transition-space-request
On Dec 1, 2011, at 8:10 AM 12/1/11, Eliot Lear wrote:
Where we ran into trouble the last time on this was that the OSS systems
themselves that manage the edge devices needed to be able to actually
communicate with those devices using the reserved space (reachability
testing, what-have-you). All that stuff runs on a variety of h/w,
including Linux, Windows, and other. But if ops want to use 240/4, I
say have at it! It's just sitting there, after all...
Got it. I mistakenly inferred you were referring back to the discussion about
adding 240.0.0.0/4 to the global address space pool...
On 12/1/11 2:06 PM, Ralph Droms wrote:
On Dec 1, 2011, at 3:35 AM 12/1/11, Eliot Lear wrote:
On 11/30/11 6:09 AM, Randy Bush wrote:
skype etc. will learn. This does prevent the breakage it just makes
it more controlled. What's the bet Skype has a patched released
within a week of this being made available?
cool. then, by that logic, let's use 240/4. the apps will patch within
a week. ok, maybe two.
As someone who tried to "Go There", I agree with you that 240/4 is not
usable. It would be fine in routers in short order, as it's fairly easy
for ISPs to exert influence and get that code changed, but general
purpose computing and all the OSS systems are a completely different
kettle of fish.
Eliot - in the case of Shared CGN space, I think all that's needed is for
the ISP routers between the CPEs and the CGN to forward 240.0.0.0/10
traffic. Those addresses will be hidden from the rest of the Internet by
the CGN on one side and the subscriber GWs on the other side. If this
address space weren't hidden, RFC 1918 space (e.g., 10.64.0.0/10) or a /10
reserved from public IPv4 space wouldn't work, either.
Those subscriber GWs would have to handle 240.0.0.0/10 traffic correctly,
and there would likely have to be some small amount of parallel RFC 1918
space in the ISP core network for servers, hosts, etc. Of course, I'm not
an operator, so I'd be happy to hear why I'm confused.
But that actually supports the notion that we need to use a different
block of address space. So does the argument that 10/8 et al are well
deployed within SPs.
You wrote also that:
and all this is aside from the pnp, skype, ... and other breakage.
and, imiho, we can screw ipv4 life support.
To keep this in the realm of the technical, perhaps you would say (a
lot) more on how you think this would break IPv4?
For the record, I'm of two minds- I hate the idea that the SPs haven't
gotten farther along on transition, and I also wonder whether a rapider
deployment of something like 6rd would be better than renumbering all
edges into this space. On the other hand, that speaks nothing about all
the content on v4 today, and that's where the pain point is.
Ietf mailing list
Ietf mailing list