Re: Reference based Routing (RbR)

[Robert Raszuk <[e-mail]>]

Hi,

> If I would insert a reference into flow label space I am afraid I would
> lose the current flow label functionality which perhaps in some cases may
> not be desired.
>
> Is someone actually using the flow label?  As far as I know it’s wholly
> unused.
>

I really don't know. Perhaps folks on the cc-ed lists could comment if
anyone is using it for anything.

RFC2473 does not give any guidance. In fact its section 6.5 recommends to
just stuff it with value of zero.

But if we can use it within encapsulated header this is great news. All
what I stated in my original note equally applies to the case where a
pointer travels in Flow Label space. And IMHO 20 bits is sufficient.


> However the exercise here is to provide a form of path steering for IPv6
> networks where transport MPLS is not enabled by design.
>
> Is that a problem that needs to be solved?
>

Well we have SRv6. We are now discussing adoption of new Routing Header
(CRH) as described in draft-bonica-6man-comp-rtg-hdr-22

If there is no problem to be solved I think we would not be having 100s if
not 1000s of messages about it and 10s of WG documents as well as few RFCs
:)



> > Last I am not sure what "messing with address space" comment really
> meant to indicate ...
>
> Changing the semantics of any part of the address space may require more
> semantic and pragmatic changes than would be necessary via other approaches.


Very true. The placement of the pointer referencing action within the last
bits of the DA address was just an example.

The core of the main idea is that instead of listing a stack of transit
nodes to traverse in each packet, let's outsource some of it to the control
plane.

Where to carry the pointer in plain IPv6 seems like an open topic. We could
carry it in:

- 20 bit flow label
- last 32 bit of the DA
- new RH (RRH - Reference Routing Header ) of fixed size and immutable
entire content

Many thx for comments !!!
R.