krstic at solarsail.hcs.harvard.edu
Thu Jul 5 15:57:05 EDT 2012
On Jul 5, 2012, at 12:47 PM, Joshua Kroll <kroll at cs.princeton.edu> wrote:
> Anyway, if you use a hardware-supported virtual machine, the overhead
> from either one should be ~0 most of the time anyway, right? That is,
> at least, how I understand hardware-supported virtualization: the only
> overhead comes from multiplexing hardware access and from running
> things in the hypervisor.
The hit is not extremely close to zero; it's detectable, but also consistent enough to let you determine a baseline and compare against that.
More information about the hcs-discuss