Re: [P2PSIP] New draft: HIP BONE

> OK, I'm not the architecture monitor, but I disagree with some of  
> this, and agree with the rest...

:-)

>> In the case of HIP BONE, the peer (or routing) protocol is both a   
>> user of HIP and an enable of HIP.  But so are the routing  
>> protocols  for IP, and in certain sense DNS for IP, too.
>
> OSPF does use IP *encapsulation*, *fragmentation/reassembly*, etc.  
> but OSPF conversations happen with direct adjacencies. Using IP  
> *encapsulation* is very different from using *IP*.

Exactly.  And that is very much how HIP would be used in HIP BONE for  
creating connectivity between peer nodes that are adjacent according  
to their finger (or whatever routing) tables.  In that case the peer  
protocol is a user of HIP for encapsulation, security, mobility,  
multi-homing, and NAT traversal, but not for directly forwarding  
signalling messages from one overlay node to another.

In the case of routing initial HIP signalling messages for creating  
direct peer-to-peer connections, HIP is then a user of the peer  
protocol, the peer protocol being an enabler for HIP.

> IS-IS runs directly over link-layer protocols ("same level as CLNS").

I know.  Did you intentionally leave out e.g. BGP which does use both  
IP and also TCP?  :-)

> DNS looks like a IP enabler to applications that are trying to use  
> DNS names, but it's not - IP ran for years before there was a  
> naming system at all (see HOSTS.TXT).
>
>> Such a circular dependency in itself is not bad, as long as it is  
>> designed (i.e. not accidental) and we have a clear understanding  
>> how  to manage the bootstrap problem.
>
> I would agree with both points here - design and understanding are  
> what's needed, and being able to bootstrap would be the first test  
> that we have a design and understanding.

I couldn't agree more.  Personally, I don't see any big problems in  
bootstrapping.  The only bigger problems I see are related to  
supporting multiple name spaces aka multiple overlays and allowing  
nodes to participate to a number of them same time.  There I see at  
least three or four approaches, and I have big difficulties in  
understanding the tradeoffs, as I don't understand the requirements  
or scenarios well enough.

--Pekka

_______________________________________________
P2PSIP mailing list
P2PSIP@ietf.org
https://www1.ietf.org/mailman/listinfo/p2psip