Today, private clouds are gaining popularity among businesses due to privacy issues and lack of control in public clouds. Most private clouds still have unique APIs which prevents users and businesses to move across different cloud platforms. Combining different private cloud platforms within a production environment may be advantageous, but is not possible with todays system without reimplementing the business logic.
The goal of this thesis is to explore and investigate this problem and create a prototype tool which makes it possible to move virtual machines across heterogeneous cloud environments. The main reason for creating such a tool, is to gain flexibility in cloud and virtualization platform choices and to prevent vendor lock-in.
The working implementation in this thesis shows that the prototype succeeds in creating a total heterogeneous environment with no dependencies, such as shared storage.
Throughout the thesis the performance was measured and analyzed to give an indication on how the expected virtual machine behavior would be. Further efficiency theories were addressed to pursue the thought of bringing different private clouds into separate environments in a coherent manner.