Re: [spring] “SRV6+” complexity in forwarding

[Robert Raszuk <[e-mail]>]

Hi Stewart,

Yes this is exactly what I was trying to say. Your definition of TI-LFA as
expressed below is spot on:

"I think a better description for the technology is that it is not
constrained by topology, i.e. that you can create the repair path
regardless of the topology, although the more perverse the topology the
more SIDs are required. "

Thx,
R.


On Fri, Sep 20, 2019 at 4:16 PM Stewart Bryant <[e-mail]>
wrote:

>
>
> On 20/09/2019 09:44, Robert Raszuk wrote:
> > TI - stands for Topology Independent ... all other LFA modes rely on
> > topologies to be able to compute or not the backup path.
>
> Well so does TI-LFA.
>
> At some level you have to know the topology to calculate *any* path in
> SR, else how do you know what links and SIDs to use. When you use loose
> source routing that is even more the case.
>
> Moreover since TI-LFA specifies the need to align the repair path with
> the post convergence path you have to know the post convergence path to
> do that, and to do that someone must know the topology.
>
> I think a better description for the technology is that it is not
> constrained by topology, i.e. that you can create the repair path
> regardless of the topology, although the more perverse the topology the
> more SIDs are required.
>
> TI-LFA is only independent of the topology it that it can find a path
> regardless of the topology, not without knowledge of the topology.
>
> - Stewart
>
>
>