[rbridge] Resolving comments in Quebec City on draft-ietf-trill-rbridge-options-05
anoop at alumni.duke.edu
Thu Sep 29 12:18:37 PDT 2011
Some comments inline.
On Wed, Sep 28, 2011 at 6:37 PM, Donald Eastlake <d3e3e3 at gmail.com> wrote:
> I'd like to go ahead and make changes in the TRILL Header Extensions
> (aka Options) draft based on my presentation at the most recent TRILL
> WG meeting in Quebec City and resolving the comments made at that
> meeting. Basically, the deviation from the changes in my presentation,
> to resolve the comments made, would be the following:
> 1) More radical simplification: There were comments that the total set
> of mechanisms in the draft seemed too complex and insufficiently
> motivated. Suggested resolution is to move the TLV options and ECN to
> a separate document to make it easier to advance the remaining
> simplified header extension draft.
This makes sense. Having multiple ways of defining options
was adding unnecessary complexity.
> 2) Alert bit: This was described as forcing the frame to the slow
> path, a description which constrains implementations unnecessarily.
> Furthermore, the motivation is the somewhat narrower case of RBridge
> Channel protocols that want hop-by-hop processing. Suggested
> resolution is to call it the RBridge Channel Alert bit and just say
> that it indicates that the frame is an RBridge Channel protocol frame
> that is requesting hop-by-hop processing. It would be left to
> implementations whether this involved diverting the frame to the slow
> path or something else.
No strong opinion on this.
> 3) Border bits: There were some bits in my presentation, that were
> labeled as being related to "border RBridges", that is, hypothetical
> multi-level IS-IS RBridges that connected to both Level 1 and Level 2.
> But such RBridges have not yet been specified. Suggested resolution is
> to just label these bits as "reserved".
Makes sense since the functionality is not yet defined.
More information about the rbridge