Showing results for 
Search instead for 
Did you mean: 

FEX FET-40G vs chassis FET-10G connections



I'm looking for an answer how to attach a 40Gbps FEX (let's say N2K-C2348TQ) to a 10Gbps Nexus chassis (let's say N77-F348XP-23 in a 7706 chassis, running 7.3(1)D1(1)).
The 7706 only has 10Gbps linecards, no 40GBASE interfaces.


I see 2 options:

  • using FET-40G in the FEX towards 4x FET-10G in the 7706 chassis.
  • using QSFP-4X10G-AOCxM breakout cables, max distance is 10m (this distance is too short to cover all FEX'es)

I'm confused on how this is interoperable...? The FEX will detect a 40GBASE interface, while on the 7706 there will be 4x 10GBASE interfaces.
I've already discovered we can convert 4 10GBASE ports into 40GBASE mode through 'breakout' cli configuration in NX-OS. This is essential/key to make this FET-40G<->FET-10G work, right?
Otherwise we'll end up with 40GBASE on 1 side (FEX) and 4x 10GBASE on the other side (7706) which cannot work.
That being said, the N77-F348XP-23 doesn't offer support for breakout cli configuration so this solution will not work on the existing linecard. I'm just trying to understand the concept of 40GBASE vs 10GBASE...


I'm also confused on how QSFP-4X10G-AOCxM would present itself towards the FEX. Will it force the FEX to split the 40Gbps into 4x 10GBASE interfaces?
This post made me understand the breakout cable would force the switch to split it's 40Gbps uplink into 4x 10Gbps uplinks:
"Once connected do "show fex 1xx transceiver" and you will see (24) FEX Uplinks available. (4) for each 40G port."

If this is correct, do we actually need to connect all 4x 10Gbps in the chassis? Or can we connect eg 2x 10Gbps in the chassis resulting in a 2x10Gbps FEX uplink? (it's because initially there might not be enough free ports on the available F3 cards)




Everyone's tags (4)
CreatePlease to create content
Content for Community-Ad