SMB random write

In Figure 1 a random-write FIO workload is running against a file system using SMB-connected shares. The workload is scaled from 1 to 128 jobs running on the KVM guest. The storage space is provided from the protocol server, contributing an additional 64 GiB of page cache.

Figure 1. SMB random write
SMB random write

Throughput peaks at 16 jobs. CPU peaks just under 2 CPUs. The throughput increases with the increasing amount of jobs, with a peak at 16 jobs; then it decreases. The CPU load shows a corresponding behavior, where the highest throughput values are at the low and high end of the amount of jobs.