Re: [DNA] Re: RS/RA Exchange

[Greg Daley <greg.daley@eng.monash.edu.au>]

Hi Brian,

Brian Haberman wrote:
> Hi Greg,
> 
> Greg Daley wrote:
> 
>> Hi Brian,
>>
>> Brian Haberman wrote:
>>
>>>
>>>
>>> Erik Nordmark wrote:
>>>
>>>>> I'd guess that the ordering and selection issue is a generic one.
>>>>> It would be good if we didn't re-invent solutions.
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> There are protocols, like MLD and routing protocols, which elect some
>>>> form of designated node (querier for MLD) but those protocols don't 
>>>> produce
>>>> a full sorted list of all the routers on the link.
>>>> I wonder if there is something here which could be reused.
>>>
>>>
>>>
>>>
>>> So, I can dust off some "historic" code for this.  DVMRP has a HELLO
>>> message like many other routing protocols.  But, DVMRP uses them to
>>> announce reachability and discovery.  That is, peer routers on a link
>>> look in the other routers' HELLO messages for their own addresses.
>>> This allows them to determine when they have been "discovered" by
>>> their peers.
>>
>>
>>
>> This sounds interesting as allows bidirectional checks that the
>> router believes in the presence of the peer.
> 
> 
> Yes.
> 
>>
>> This means that at least the discovery mechanism is assured.
>> If we can do this based on existing message exchanges or
>> mechanisms (or code :) that would be good though.
> 
> 
> How about an RA option that lists all "learned" peers?

I've just proposed another mechanism for ordering in
a response to Erik's mail, which only needs one or two
identifiers (hashes of lists of routers).

If we determined the requirements for such a system
(roughly), we could see if the simpler mechanism
including the router list is sufficient.

This may be the case if we only need three fast routers
(for example) and for other routers to know they're not
fast (and use RFC2461 delays).

>>
>>> Could we mimic this and have some type of announcement message that
>>> a router would send out with all the other routers' addresses that
>>> it has heard from?
>>
>>
>>
>> I thought that there was some work going on on
>> a routing protocol agnostic 'hello' mechanism.
>>
> 
> The agnostic HELLO work is meant to have fast (re: millisecond)
> detection of neighbor loss.  One proposal is PLP, there was one
> a few years back called FLIPP.  Not sure if that is really the
> model we want here.

OK.  Probably it's a bit pie-in-the-sky for the moment.

RS/RA exchange may be sufficient, if we have options
which can be used.

Greg