[rbridge] Threat and security analysis?
erik.nordmark at oracle.com
Fri May 28 10:28:25 PDT 2010
On 05/28/10 08:44 AM, Jukka Manner wrote:
> Hi Joe,
> I was mostly thinking in the direction of
> a) Deployment guidelines: what people should understand when looking for
> deploying TRILL. Yes, concrete details would be good.
> b) Future work: how could we make TRILL more secure and trustworthy than
> legacy Ethernet since we don't need to carry all that legacy (yes, I
> know, the simplicity requirement of TRILL is against this).
Earlier we had talked about the option of getting more security using
the combination of:
- IS-IS with reasonable security (something which already exists in IS-IS)
- Using ESADI for host address announcements (supported in the
standard with higher confidence then the learned addresses)
- Using 802.1X or similar as the only want an edge rbridge would add a
MAC address to its ESADI announcements.
If somebody wants to work on this it might make sense to right it up as
an informational document. I don't think any new standards work would be
needed - it is more a question about gluing together the pieces and
looking at what the resulting security would be for such an approach.
More information about the rbridge