Nexus 9000-series chassis / CPU routing

[tham89]

It's been found that there was formerly a strong recommendation for the Nexus 7K-series switches (on page 27, 40), where under the following conditions:
- Point-to-point routed links, terminating on a vPC domain
- "Backup path routing" needed using l3 peer-gateway enhancement

that recommendation is that Engineers not use an SVI across the peer-link to establish the backup path since that would have been CPU-bound.

Does anyone have experience, or inside knowledge, that these requirements which are mysteriously missing from all of the current 9K-series switches guides have really vanished, and routing of this sort is handled properly by the ASICs? If it makes any difference, I'm using 97-series Cloud-scale Line cards.

I've carefully reviewed the following, and don't see any evidence that this is still a special case, but I can admit that I'm wrong in the face of overwhelming evidence:

Cisco Nexus 9000 Series NX-OS Interfaces Configuration Guide, Release 9.3(x) - Configuring vPCs

Understand and Configure Nexus 9000 vPC with Best Practices

So,... I'm not sure!

Thanks!

[tham89]

... any takers?

[Sergiu.Daniluk]

Nexus 7K and Nexus 9K have completely different architecture when it comes to hardware forwarding, since there are different ASICs on them. And I am not saying different generations. No, it's actually different architecture. Same for N5K/N6K. Completely different then N7k and N9k. So if there are requirements or recommendations of any kind for N7K, it is not implied that it is also applicable for N9K. You should always check and follow the config guide and release notes of the specific platform/NXOS version you have.

Take care,

Sergiu