Problem optimization is a fundamental task encountered everywhere, from everydays life to the most complex science areas. Finding the optimal solution often takes an unreasonable amount of time or computing resources. Therefore, approximation techniques are used to find near-optimal solutions. Bio-inspired algorithms provide such approximation techniques, they mimic existing solutions found in the nature. But even those techniques are sometimes to slow for extensive problems, so they need to be run in parallel.
This master thesis presents a new framework, Biohadoop, to facilitate the implementation and execution of parallelized bio-inspired optimization techniques on Apache Hadoop. Its usefulness is demonstrated by the implementation and performance evaluation of two bio-inspired optimization algorithms.