FLEX Link Alternate on Catalyst 9K

[Taandav21]

Hello There,

Although I am under the impression that catalyst 9K series doesn't support FLEX link, if someone knows otherwise, please correct me. Also, if my impression is correct, is there any alternate for FLEX link on Cat 9Ks? Or RSTP is the way to go? Looking to upgrade the existing 2960X switches but FLEX link is a needed feature. 

Thank you,

TT

[paul driver]

Hello

the alternative to flex on cat 9000’s is object tracking (ipsla- line-protocol - route etc) 

[Taandav21]

Thanks, Paul. I was wondering, whats the best convergence time we can achieve with IP SLA/Object Tracking. Would it be faster than RSTP? RSTP, in general, converges around 8 seconds, as per some documents. FLEX link converges much faster than that. As per a document on this forum, it converges within less than 50 msec.

https://community.cisco.com/t5/networking-blogs/flexlink-in-cisco-layer-2-network/ba-p/3106779

Idea is to reduce the convergence time. Any other options?

[paul driver]

Hello

rstp and ipsla object tracking are two different features - 

rstp is a l2 feature 

object tracking ipsla is L3 

Depending on what you are wanting to do you also have other options for dynamic routing convergence one such feature Bidirectional Forwarding Detection (BFD) 

Can you elaborate on what you want to achieve?

[Taandav21]

I understand that. It may not be possible to deploy IP SLA. I was thinking, REP may be an option?

[paul driver]

Hello

i honestly cannot comment on REP I am aware of the feature and know it has limited functionality as apposed to STP but as far as that never used or come across it in a production network - maybe others on this forum have more experience and exposure to it to comment and advise.

As far a convergence I assume by the above you are looking at a L2 convergence correct?

[Taandav21]

Thats fine, Paul. I am in the same boat. Your assumption is spot on. Its a typical Access and Collapsed core design. Fast L2 convergence is the objective. Read a white paper and it seems, REP has sub-second convergence times. To be specific, 50-250 ms. Although its meant for metro ethernet rings, as per the document, it can be suited for other topologies as well.

It would be interesting to find out, if someone has actually implemented REP in a campus LAN environment.

In any case, thank you for your help.