Re: [Hipsec] Looking for slides on Relay server

[Miika Komu <[e-mail]>]

Hi,

On 05/22/2014 11:50 PM, Robert Moskowitz wrote:
>
> On 05/22/2014 04:36 PM, Miika Komu wrote:
>> Hi,
>>
>> On 05/22/2014 11:16 PM, Robert Moskowitz wrote:
>>> At times I would like to strangle myself. WHY did I ever create private
>>> addresses for IPv4 and thus create a market for NAT boxes????? Well if
>>> I have not been involved, it would have still happened. The use cases
>>> were out there and ROAD was dead. Enough handwringing. We have Nasty
>>> NATs and mobile devices pop in and out of them. So we have to relay.
>>>
>>> But only the WiFi connection would get behind bad NATs. My testing over
>>> Verizon Wireless has worked well without relaying. So one MIGHT think
>>> that with LOCATORs we could say that this locator need not relay, but
>>> this better. Of course the phones have this tendency to roam and
>>> perhaps not all cellular providers are set up not to need relaying....
>>>
>>> Andrei's lecture notes do not cover the relay server part, only rvs and
>>> I3 stuff. I am looking for some slides to cover relay.
>>
>> in HIPL project, we have been just using the Teredo infrastructure
>> (with miredo software in Linux) to take care of the NAT penetration
>> and HIP for its persistent namespace (as Teredo addresses can change).
>
> And when you do Teredo, you only need RVS, not Relay. Actually, the
> Relaying is done in Teredo.

You still need RVS for double jump.

> My head hurts. But this means you are doing Teredo NAT traversal not HIP
> NAT traversal?

Currently, HIP for Linux supports just full relay. The bis-style NAT 
traversal extensions are not supported. We did implement RFC5770, but 
this was removed from the trunk, so you have to use Teredo if you don't 
like the full relay. Anyway, this is getting a bit implementation 
specific for this mailing list.