Parallelizing data stream management system with OpenMP: A case study

Parallelism has been a hot topic for decades in computer science community. There has been huge demand of parallelizing serial codes for multicore CPU environment. This paper presents parallelism issues, discuss parallelism strategies and present empirical performance results. The study was conducted on AQSIOS: Advanced Query System Infrastructure on Streams for enhancing the performance of the software. At the time of peak load, even though system had multiple CPUs with multi cores, only single CPU core was used. It was then decided by authors to parallelize the code so that speedup can be achieved. We chose OpenMP programming system to attain parallelism. Our results of experiments shows that considerable speedup can be gained by exploiting parallelism on various parts of the DSMS.