[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: ND-proxy applicability and loop-prevention



Hi Margaret,

Margaret Wasserman wrote:
> At 12:56 PM +0200 3/23/04, Jari Arkko wrote:
> 
>>>  I think it would be possible to detect loops if we wanted really 
>>> hard to do that.  That might just
>>>  lead to reinvention of the spanning tree protocol, though.
>>
>>
>> This is a danger, yes.
> 
> 
> Why is this a danger?

I think there's only a danger if we end up doing more work than
we need to, or work on something covered by other standards
bodies.

> IMO, the ND proxy mechanism effectively does bridging at layer 3. What 
> makes you think it will be any easier here than at layer 2?  If layer 3 
> bridging is something that we need, then why not run STP?

I think that using the spanning tree algorithm may be OK,
if we can work out appropriate root brouter selection, metrics
etc.

I'd suggest that extending the coverage of a bridged network with
802.1d/w spanning tree protocols is not a good idea though,
for a number fo reasons.

Firstly, the computational/message complexity of STP is such that
the convergence times will increase greatly by connecting
disparate bridged networks. (Even just running a separate
spanning tree amongst the proxies for a level of hierarchy would
work better, as area 0 does in OSPF).

Secondly, there are several issues in 802.1d/w networks
which are constraints due to the transparent nature of the
bridging in these networks (no modification to L2 frames).

In a proxied network, we can get much more control of the forwarding
plane by changing the L2 headers.
This may allow us to avoid the lack of redundancy present
in spanning trees (for example using DAGs or shortest paths
instead for unicast forwarding).

So L3 spanning trees may be one way to create a routing topology
for ND/MLD proxies, but it may not be the best way.

> There was some discussion of this general concept (zero-configuration 
> routing and/or layer 3 bridging) in the Zerouter BOF and on the zerouter 
> mailing list...  Does anyone know if that effort is still active?

I'm not sure where zerouter work is at the moment.

The bof chair (Aidan Williams) was working at motorola
labs in Sydney which has closed down. I've had difficulty
tracking him down (mail bounces).

Radia Perlman and Aidan were working on this before.
in a (now expired) draft.

http://www.watersprings.org/pub/id/draft-perlman-zerouter-rbridge-00.txt


Greg


--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------