We know that when you’re conducting valuable research, cluster management can be a time-consuming, frustrating process, particularly if you don’t have a background in hardware setup.
We designed Cluster in a Box to combat this problem. Built with research groups and university spin outs in mind, this HPC cluster acts as a foundation for those looking to make the most of their existing and future compute. It offers all you need to get started and can be easily scaled as your team and research expand.
In some areas, we’ve deviated from the parts and software you’d usually find in a HPC cluster – instead opting for innovative and open-source components to keep things cost-effective.
SupremeRAID™ SR-1000
If you aren’t already familiar, SupremeRAID™ utilises GPU-accelerated storage to remove bottlenecks traditionally associated with RAID, delivering maximum SSD performance without creating throughput bottlenecks or consuming CPU cycles.
We chose this component to increase the efficiency of your storage without increasing costs. SupremeRAID™ cards offer incredibly high performance and flexibility, thanks to their innovative out-of-path technology.
Slurm Workload Manager
Slurm Workload Manager is a fault-tolerant cluster management and job scheduling system built to manage Linux clusters. Its core functions involve allocating resources to jobs; providing a framework for starting, executing and monitoring work on these resources; and managing a queue of pending jobs to arbitrate contention for resources. This works to greatly maximise the efficiency of your cluster’s jobs, saving you time and money.
Slurm was our job scheduling software of choice due to its high configurability and fault tolerance, making it optimal for HPC research. It’s also open source, so you’ll benefit from extensive documentation and community support, and pay a much lower cost for the privilege.
Proxmox Virtual Environment
After recent price rises in the VM space, we wanted to opt for software which would not only cut your upfront costs, but would also save you from potential price hikes in future.
Proxmox VE provides a single interface to manage virtual machines, software-defined storage and more, to streamline your cluster management. It’s open source, meaning it’s substantially cheaper than traditional, more popular alternatives. Like with Slurm, you’ll also have access to beneficial community support and documentation, as well as enterprise-grade support services and training at a much lower price point.
Whilst there are other VM options, such as Nutanix, we opted for Proxmox because its cost-effectiveness, flexibility and ease of use make it more beneficial to smaller organisations, like spinouts and research groups.
Overall, Cluster in a Box was carefully curated to scale with you and your research, to boost your efficiency and keep your costs low.
Ultra high-speed storage is paired with efficient job scheduling and virtualisation software to boost the efficiency of your workloads, saving you valuable time and costs. Role-based access controls, image management and shared storage provide a collaborative framework for your team to work from, enhancing your working environment.
Our software stack is open-source and mostly license-free, keeping your costs low and offering extensive documentation and community support. More efficient energy use will also cut your expenses, reduce your carbon emissions and help you meet ESG regulations.
If you’d like to learn more, get in touch with our experts below to learn more or start configuring your own cluster.
