your company has a monthly big data workload, running for about 2 hours, which can be efficiently distributed across multiple servers of various sizes, with a variable number of cpus. the solution for the workload should be able to withstand server failures.
