This is possibly too late, but I'd personally love to see how it would handle two concurrent Juniper vMX
"instances" (or 4 vQFX-10K
instances) either running in Linux using KVM, or in ESXi. I've got instances in quotes, since Juniper split both vMX and vQFX into two separate VMs that you need to interconnect. A routing engine VM, and a packet forwarding engine VM. I wish it could be three vMX instances, but the minimum requirements for the lab setup is 4 "vCPUs" (h/w threads) and 6GB RAM per-instance, so that 16GB RAM will prevent that (same with using Cisco XRv 9K image. that needs 16GB per-instance). vQFX is a little more forgiving, just needing 3 h/w threads and 3GB RAM per instance, but even when I've increased the resource allocations, it still takes about 20-30 minutes to fully load up.
I've got an old i7-2600K with 32GB RAM, but those 8 h/w threads are the limiting factor, when trying to test out some EVPN w/VXLAN setups I'm wanting to try. The old vMX engineering pre-releases were a single VM that could run with just 1GB RAM, and 1 h/w thread, but it's missing a lot of functionality. Works fine as purely L3 device, though.