Linux HPC clusters provide an excellent way to harness the computing power of commodity X86-compatible rack based server and blade servers. A cluster is often composed of a number of compute nodes with a set of dedicated machines for management, storage and interactive services. A high-performance interconnect is usually included providing high-bandwidth, low-latency communication between the cluster nodes.

High Performance Computing clusters can run a wide variety of software, packaged into a collection called the HPC cluster software stack. Your choice of stack will influence the applications your cluster can provide, the hardware on which it is built and the overall cost of the solution. Common components of a stack include:

A number of pre-defined commercial HPC cluster stacks are available to help install a known-good set of software components suitable for creating a basic, general purpose HPC cluster. Alces Software can also provide our Open-Source HPC stack, combining the best freely available products with enterprise class support services. However, to realize its full potential the software stack must be carefully configured specifically for the target hardware - only when each individual component is properly tuned will maximum performance be achieved.

Alces Software specialize in advising on software stack components to allow customers to create a tailor made solution from the best commercial vendors and open-source projects. We can install, configure and demonstrate the cluster before running acceptance tests to confirm proper operation of the hardware and software stack. All our processes are fully customizable allowing you maximum control over the system delivery.

Contact us for more information on engaging Alces Software integration services.