Random workloads

Measurements were obtained for the running of a random read and random write workload while scaling the number of workload jobs. The workload was run directly on the KVM guest. These results were observed:

• The random-read, low-workload counts show closely clustered throughput values, with SMB leading in CPU efficiency. At medium workload counts, iSCSI displays a few leading values for throughput, while NFS exhibits a few leading values for CPU efficiency. Higher workload counts show SMB with leading behavior in throughput and CPU efficiency.
• The random write shows SMB with peak throughput for low, medium, and high workload counts. SMB exhibits leading CPU efficiency for low and medium job counts, with NFS displaying leading CPU efficiency in a few medium workload counts.