I have a fairly old Dell R720 server, running the neat multi-purpose Unraid OS. It’s primarily a file/media/whatever server. It’s hosting this webpage you’re reading. It’s also got a 30 core, 100 GiB RAM VM running eve-ng on it. For ages (as part of my procrastination to get a new cert, like finally pushing my JNCIP-SP into a JNCIE…) I have had eve-ng running with a mix of vMX and vSRX nodes. Due to the way vMX and vSRX work, they smash your CPU cores. This might be fine in a production environment, but for a lab designed to play with protocols - it’s overkill. This is especially bad when combined with running in a nested VM, sharing the server with more important things (like media serving…).

So - Juniper released a virtual route reflector, designed to only be a route reflector - running without DPDK CPU smashing. The result for me, was being able to run a 12 node lab, based on the MPLS in the SDN Era book - with a total CPU toll of around 20%. This is going down from constant 100% when using only a couple of vSRXes. Incredible, and just what I wanted. Quite similar to the ‘Olive’ SRX of yore.

vRR is available from Juniper’s website the same way you can grab their other images (with an account and correct permissions).

Does the vRR do everything I want for my advanced MPLS lab work? Can it handle the EVPN jandal required for a JNCIE-SP in 2023? I will keep you posted. Overall, if you’re looking for a virtualised Juniper lab, vRR might be the way to go.

Shoutouts to Adam (@packetsource) for reminding me vRR exists!