I was wondering, what is the purpose of a dual-port NIC on a blade compute unit, in the context of a datacenter rack, like which of these configurations is more in line with modern datacenter architecture?

  1. Having a backup top-of-rack switch (is this normal?) and connecting one port of my NIC to the regular one and one to the backup.
  2. My compute unit is NUMA dual socket (so 28 cores on one, 28 on the other socket). Reserving one port for traffic generated by/accepted by each socket? (Can we have top-of-rack switch redundancy with this config?)
  3. Or something else?

Just to provide context, I am trying to figure out how many partitions my rack can support. I am using a Tofino2 switch with the equivalent of 128 100GbE ports, each of my compute nodes is dual-socket 28-core per socket, and my NICs are dual-port 100GbE. So if each NIC only hooked up one port to the primary switch as in option (1), there would be 2x scale out, double than in option (2).

Thank you!

1 Answer 1


There are many use cases for dual NICs, including

  • redundant active/passive failover, usually with redundant switches as well
  • active/active load balancing
  • split, dedicated use by different applications, VMs or networks (LAN/SAN, WAN/LAN, production/management, ...)
  • any combination of above

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.