I'm having a Hyper-V hypervisor and am running a CentOS 7 VM on top of it. The VM houses certain microbenchmarks (sysbench, stream, iozone and netperf) and I'm trying to study the effects (utilisation) of resources (CPU, memory, disk, network bandwidth) for each.
To study resource utilisation, I'm using node exporter and prometheus. Is there any way that I can split the usage? For example, if a compute-intensive task (sysbench) is running in a VM, then what percentage of CPU cycles are used in executing the VM's (guest) code vs those used in executing the hypervisor's code
Any help in this regard would be greatly appreciated