Analysis of Entropy Levels in the Entropy Pool of Random Number Generator

The term entropy is in computer science usualy used to refer to a stream of random data. This work sumarizes briefly techniques used to generate random data and describes the random number generator used in the Linux kernel. Later on this work focuses on determining the bit generation speed of the Linux kernel RNG when running as virtual machines under different hypervisors. The work describes the reasons for poor performance of the RNG in virtual environment and proposes steps to overcome them. As a next step, the proposed approach is implemented, tested and the results are compared with the original system. The entropy distribution system is able to improve the level of entropy in the kernel by orders of magnitude when using a fast RNG as a source.