On low-power SoCs as storage bricks for Bioinformatics

In the perspective of energy-efficient storage solutions supporting scientific computing, we investigate the possibility of using low-power Systems-On-Chip (SoCs) as storage bricks of a BeeGFS file system in support of a realistic genome sequencing pipeline from Bioinformatics. Metadata and data performances of such a file system made of low-power devices are assessed via various benchmarks for an increasing number of clients accessing the data, and compared to those achieved with traditional power-hungry architectures with a similar BeeGFS setup. Then, we show how the sequencing information streamed by portable sequencing devices, such as the Oxford Nanopore MinION, can be managed by the underlying BeeGFS file system built with low-power SoCs.