Alejandro Duarte’s Fast Cluster Packs Eight Raspberry Pis for 32 Cores and 64GB of RAM



Developer Alejandro Duarte has put collectively a low-power cluster for database experimentation, strapping eight Raspberry Pi 4 Mannequin B single-board computer systems collectively for 32 processor cores and a whopping 64GB of RAM in complete.

“Since I began to find MariaDB and discovered about database clusters, Distributed SQL, and Xpand, the thought of constructing a Raspberry Pi cluster has been behind my head,” Duarte explains of the venture’s inspiration. “A cluster like it is a nice strategy to experiment with distributed techniques.”

Whereas Duarte factors out which you can experiment with the identical applied sciences on a single Raspberry Pi, his construct makes use of eight Raspberry Pi 4 Mannequin B 8GB single-board computer systems — the best specification model accessible on the time of writing — linked to a ten-port 60W USB charger. With every Raspberry Pi having 4 processor cores, that is a complete of 32 cores and 64GB of RAM — whereas a 256GB microSD card in every offers a complete of 2TB of storage.

The cluster runs on Raspberry Pi OS Lite, put in for headless operation. The {hardware} is stacked eight-tall utilizing stand-offs and acrylic sheets, with two followers included to assist hold issues cool. For post-installation configuration and cluster management, Duarte recommends Ansible — permitting for purposes to be put in and executed throughout all nodes of the cluster.

“You now have a cluster of small computer systems able to be managed by Ansible. You may strive organising any form of server software program, together with net servers or databases,” Duarte writes. “There is also way more about Ansible. I merely scratched the floor right here utilizing ad-hoc instructions, however you’ll be able to for instance create playbooks that comprise the state wherein you need your cluster to be.”

See also  IoT sensors: A means of sensible automating agriculture

Duarte’s full information is on the market on DZone.

Leave a Reply