RE: RNET: Random Network Endpoint Technology

> -----Original Message-----
> From: Chad Giffin [mailto:typosity@hotmail.com] 
> Sent: Saturday, June 21, 2008 10:53 AM
> To: Dan Wing
> Cc: IETF
> Subject: RE: RNET: Random Network Endpoint Technology
> 
> > From: dwing@cisco.com
> > To: typosity@hotmail.com; ietf@ietf.org
> > Subject: RE: RNET: Randon Network Endpoint Technology
> > Date: Thu, 19 Jun 2008 09:57:18 -0700
> >
> >> -----Original Message-----
> >> From: ietf-bounces@ietf.org [mailto:ietf-bounces@ietf.org] On
> >> Behalf Of Chad Giffin
> >> Sent: Thursday, June 19, 2008 9:49 AM
> >> To: IETF
> >> Subject: RNET: Randon Network Endpoint Technology
> >>
> >> June 18th, 1145h CDT
> >>
> >> To all members of the IETF mailing list;
> >>
> >> I have posted a description, twice, of the RNET protocol
> >> to this mailing list. I have also provided some updates
> >> concerning peer to peer connections between RNET Hosts.
> >>
> >> I have yet to receive /any/ response (other then an
> >> email with an empty body) concerning by postings.
> >
> > Here is a response, which appeared to have been CC'd to you:
> > http://www.ietf.org/mail-archive/web/ietf/current/msg51774.html
> 
> This message was actually posted by me :-)

It was posted by Eric Burger.  He wrote:

  *> From: eburger at standardstrack.com
  *> To: "Chad Giffin" <typosity at hotmail.com>, ietf at ietf.org
  *> Date: Wed, 21 May 2008 21:49:24 +0000
  *>
  *> So we have reinvented STUN?
  ...

> > I agree with Eric; based on the description of RNET, it 
> > sounds much like STUN
> > combined with a rendezvous protocol (e.g., SIP). RNET is 
> > also similar to HIP's NAT traversal.
> >
> > STUN is RFC3489 and draft-ietf-behave-rfc3489bis. SIP is 
> > RFC3261. The use of
> > STUN with SIP is best described in 
> > draft-ietf-sipping-nat-scenarios. HIP's
> > NAT traversal is described in draft-ietf-hip-nat-traversal.
> 
> I looked, albeit briefly, at STUN and SIP.  these protocols 
> are not at all like what I am suggesting.
>
> RNET will punch through firewalls/NATs without a problem.   
> Peer to Peer communication using RNET Host Addresses, 
> however, may present a problem when there are NATs between 
> them.  (The answer to this is simply to allow authenticated 
> RNET Route Requests to be made at every NAT/firewall)

Incoming messages are not just an authorization concern for
a NAPT -- more importantly, the NAPT needs to know where to route 
an incoming message.  For example, if there are two RNET-capable 
hosts behind a NAPT and an RNET message arrives on the NAPT's
public interface (that is, it arrives from the Internet), the 
NAPT will not know which RNET-capable host should get the 
message.  NAPTs resolve this delimma by first expecting (and
requiring) a packet to be sent by the 'inside' host to the
'outside' (Internet) host.

> I think you missed the point of RNET. 

That is likely true.

> The point being that 
> you have a valid IPv6 IP address and are able to plug into 
> any part of the internet and use it from that location.   

That sounds like Teredo, 
http://en.wikipedia.org/wiki/Teredo_tunneling

> Your address is NOT advertised.  The routes made for 
> communication by your RNET Host decay so as not to polute the 
> internet's routing tables.
> 
> RNET is quite simple, easy to impliment.
> 
> RNET Route Requests and RNET Error Messages can be put 
> together under a new IP protocol, named RNET.  All that needs 
> to be done is to have a new protocol number assigned for this purpose.

It is not possible to deploy a new protocol behind a NAPT -- a NAPT
only understands how to translate UDP, TCP, ICMP, and (if enabled)
IPSec ESP.  RNET would have to be tunneled over UDP to be deployed
beyond a NAPT -- unless your goal is to have everyone upgrade their
NAPTs to RNET-aware NAPT devices.

-d

_______________________________________________
IETF mailing list
IETF@ietf.org
https://www.ietf.org/mailman/listinfo/ietf