Let us be clear on what a virtual particle is:It is a mathematical expression within an integral which when evaluated will give the probability or the lifetime of an event happening at (x,y,z,t), a crossection for the process. This is schematically represented with Feynman diagrams. Below are two diagrams that exchange a charged W meson which as a particle has a mass close to 100 GeV.
These are to first order in the expansion for calculating the crossections and the decay.
The crossection can be calculated from low input energies, much lower than the mass of the virtual W. The second diagram on the right gives the decay of the muon, which has a mass of 105 MeV, to compare with a real W which is around 100GeV. The line represents the quantum numbers of the particle that has to be exchanged, put the mass evidently has to be very off mass shell. The mathematical representation , the propagator, has the mass of the W in the denominator but the energy momentum variables of the interaction are way off when the integral is taken, as energy and momentum have to be conserved . A virtual particle is not the particle, it just carries the quantum labels.
Since virtual particles have charge and motion, their fields should interact with each other to produce Lorentz forces, even in a perfect vacuum.
A second lesson to take from these diagrams is that the interaction space, the vacuum you envisage in your question, is very small much less than a Fermi ( a unit appropriate for strong interactions). A virtual W- will not be found in the vacuum of space except within very very short distances under the integral for the interaction.
The virtual photon diagrams under special circumstances can be considered over macroscopic distances when talking of the repulsion of two electrons in vacuum, again a mathematical story but its effect is seen macroscopically in the electric field of the pair.