Currently we can observe the growing importance of distributed systems and algorithms in the world of IT. Their application is getting more and more widespread not just in the corporate sector (enterprise world) but also in daily used services like cloud computing, Skype or P2P file sharing. The design and implementation of these algorithms need special knowledge and expertise. The development process should also be supported by special development and test tools.
In the first part of my thesis, I will introduce the distributed algorithms, especially the ones for distributed object handling. Beyond the general description, I will also mention the common difficulties and practices of their development.
It is a common problem during the development of a distributed system, that the aggregated effects of the numerous parameters cannot be calculated or predicted properly. In this case the real behavior of the system can only be checked after the implementation. This makes the debugging and tuning of these algorithms much more difficult than it would be in design-time.
To solve this problem I will design and implement a simulation software in the second part of the thesis. The created framework will enable us to observe, test and measure distributed algorithms without the implementation of the full distributed system.
I will prove the correctness of the functionality of the simulator by applying empirical tests. After that, I will simulate and analyze some of the object access algorithm variations presented in the first part of the thesis.