Experimental verification of Hyper-V performance isolation quality level

The need for cost optimization in a broad sense constitutes the basis of operation of every enterprise. In the case of IT structure (which is present in almost every field of activity these days), one of the most commonly applied technologies leading to good cost-to-profit adjustment is virtualization. This allows the location of several operational systems with IT systems on a single server. In order for such an optimization to be carried out correctly, it has to be strictly controlled by means of allocating access to resources (which is known as performance isolation). Modern virtualizers allow us to set up this allocation in quantitative terms (the number of processors, size of RAM, or disc space). It appears, however, that in qualitative terms (processor time, RAM, or hard disc bandwidth), the actual allocation of resources does not always correspond with this configuration. This paper provides an experimental presentation of the achievable quality level of performance isolation of the Hyper-V virtualizer.