A Custom Cluster Design With Raspberry Pi for Parallel Programming and Deployment of Private Cloud

• Published in: Role of Edge Analytics in Sustainable Smart City Development pp. 273 - 288
• Main Authors: Sukesh, B, Venkatesh, K, Srinivas, L.N.B
• Format: Book Chapter
• Language: English
• Published: United States John Wiley & Sons, Incorporated 2020; John Wiley & Sons, Inc
• Subjects: master node; MPI; MPI4PY; MPICH; Nextcloud; RPi Cluster; slave node
• ISBN: 9781119681281; 1119681286
• DOI: 10.1002/9781119681328.ch14

Super Computers build on the principle of combining group of computers to provide more processing power than one computer can alone. The domain of parallel computing and supercomputing present challenges both for today and future technologies, but building a cluster requires expensive and bulk hardware such as pc's or the implementation of virtual machines. Power consumption is one of the major constraints used in building supercomputing clusters. Low power embedded processors are used in large clusters in reducing power usage rather than use of high‐end server CPUs. Single Board computer such as Raspberry pi has conquered several areas such as IOT, parallel and distributed computing within a short span. The design of cluster computing with RPi and its operations are important to the future enhancements of computer science. In this chapter, design and implementation of a cluster with 1 master node and 3 slave nodes is reported. MPICH and MPI4PY are used for parallel programming analysed speed of computation on each node. This chapter also shows an approach where we deploy one of the private clouds called as Nextcloud on RPi. A user who is using Nextcloud on his private server has full control over his data. One other approach for performance and workload distribution is that we can implement RPi Cluster where any number of Rip's can be interconnected to form a cluster. As this works with low cost and gives high performance these are termed as HPC Clusters. These RPi clusters can also replicate data centre features. We can use these clusters for hosting web services, media centre etc.