Raspberry Pi 2 as an Feasible Alternative for Cloud Based Parallel Computing Solutions

Data centres use about 250 350 TWh of electric energy per year. About 33% of the data centres power consumption comes from IT equipment. ARM devices are 3 to 4 times more efficient than the traditional x86 based devices [5]. In recent years, ARM processors have been used in small devices such as the Raspberry Pi [23]. The next generation, the Raspberry Pi 2 model B, has a higher clocked quad-core CPU. In this paper, the Raspberry Pi 2 is presented as an alternative to the execution of parallel computations on cloud computing platforms. A micro data centre consisting of 9 Raspberry Pi 2’s is build and analysed in performance and cost by execution of a RSA factorization algorithm. A comparison with two cloud systems, Amazon EC2 and Digital Ocean, is made. Also the behaviour of these Raspberry Pi’s is analysed with the aid of a model for power consumption and performance in data centres [20]. It becomes clear that the Raspberry Pi 2 is a energy efficient computer, but the performance of the Raspberry Pi in comparison with cloud systems is lacking. So large scale usage of the Raspberry Pi 2 for CPU computations not feasible at this moment compared to the cheaper, better performing, cloud systems.