Performance Impact of Thread Pooling in DSOCARE

This paper presents the results of a case study research conducted by the author that investigated the performance impact of thread pooling in dynamic selection of optimal cryptographic algorithms in a runtime environment (DSOCARE), a cryptography framework proposed in the author´s Ph.D. thesis. Using the advice, default, and pass-through priority modes in runtime optimal cryptographic algorithm selector (ROCAS) module of DSOCARE, eight cryptographic service requests with high and low security, speed, and priority quality of service (QoS) parameters were performed with and without thread pooling. The results of the performance test data showed that thread pooling increased the performance of cryptographic services by 70% in advice and pass-through priority modes. In the default mode, there was a throughput increase of 28%. The experiment also proved that the optimal number of threads in a thread pool is dependent on the number of processors and number of execution pipelines in each processor