Abstract:

This paper examines the performance database benchmarks that are available for both traditional relational database systems and Big Data technologies and ecosystems like Hadoop. It analyses the performance benchmarking in today’s database world where stakeholders want to make the choice between various data storage and processing alternatives. It argues the idea that all performance benchmarks frameworks lack randomness and hence they are not statistically relevant. Finally, it advances the idea of a “true” random performance benchmark that must be developed.