Sunday, October 25, 2020

Question about core FRR

I'm struggling to understand and choose what design is better. So there is an ISIS FRR enabled in the core and it's same switchover speed as MPLS FRR (~25-30ms). But LFA doens't cover everything due to topology. So we using TE tunnels in hot-standby (path protection) at key points where LFA doesn't work.

I've done tests to measure switchover time, so all of them have a switchover time <50ms.

So what's better? Using link-node protection (MPLS auto-frr), which creates a lot of auto-tunnels and makes troubleshooting a little bit harsh.

Or using path-protection with bfd (hot-standby), which also doubles the number of TE tunnels and requires bfd with aggressive timers?

I've read a lot of documentation and some vendors reccomended to use those both options.

Can someone share their experience with core FRR design? Maybe i'm just overthinking it, but i want to find proper solution.



No comments:

Post a Comment