[rbridge] TTL only - was RE: New fields in shim header?

[Caitlin Bestler]

rbridge-bounces at postel.org wrote:
> Ed Bowen wrote:
> 
>> What is the operational experience that indicates the requirement for
>> the other environments?  I am not aware of anyone who plans to
>> install TRILL outside of the environment Silvano outlines. What need
>> would motivate these additional environments?
> 
> Here is one data point, which isn't central to TRILL but
> illustrate the possibilities.
> 
> Barnard Aboba made the observation a long time ago that using
> TRILL in wireless access points would solve certain existing issues.
> 
> By advertising the MAC address in the link-state routing
> protocol when the station has associated, then moving from
> one AP to another is less disruptive. Today one has to rely
> on the station transmitting packets via the new AP in order
> for the learning bridges to find update the learning tables.
> 

The same advantages will apply to live migration of virtual machines
within a subnet. Indeed, a believe virtualization may be a major
factor in the need for RBridges. VMs mean more MAC addresses per
physical host, and the desire to support live migration increases
the desire for larger subnets.