Benchmarking 10GbE NAS-to-NAS Transfers: iperf, NFS, cp, and rsync
One of our Synology NAS systems was nearly full, so I planned to move data to another NAS over our existing 10GbE network. Before migrating anything, I wanted to establish realistic file-copy performance without modifying the NAS configuration. iperf2 reached 9.42 Gbps , but the copy commands told a different story: cp completed a large-file copy at 479–494 MB/s , while a single rsync process…
On November 9–10, 2026, a series of performance tests were conducted on two Synology NAS systems connected via a 10Gbps network. The objective was to benchmark different file-copying tools and understand their impact on file-transfer speeds.
The tests involved copying a 15.17 GB file (16,285,199,445 bytes) between the two NAS systems using several methods: iperf for network throughput, dd for raw data copying with and without fsync, cp for the built-in Synology copy command, and rsync for a remote file-copying utility.
iperf measurements showed that the network could sustain around 9.42 Gbps of throughput. However, actual file-copying speeds were significantly lower when using cp (494 MB/s) or rsync (181 MB/s). The discrepancy between network bandwidth and file-copying performance was attributed to additional processing overhead introduced by the file-copying tools, such as checksum computation, data fragmentation, and filesystem-specific operations.
A parallel copy test using two rsync processes demonstrated that splitting the data across multiple output files resulted in a 55% increase in aggregate throughput (281 MB/s), compared to a single rsync process. However, this method also increased CPU usage and resulted in longer individual copy times.
The CPU and RAID activity measurements revealed that cp generated the most RAID write traffic and consumed more CPU time in kernel mode, while rsync utilized more user-level processing time. These factors likely contributed to the slower single-process rsync performance.
In conclusion, the test results indicated that the choice of file-copying tool significantly impacts the overall performance, even when the network infrastructure is capable of high-speed data transfer. While iperf demonstrated the network's theoretical throughput, actual file-copying speeds were limited by the overhead introduced by the chosen utility.
The findings underscore the importance of selecting appropriate tools and understanding their underlying mechanisms when planning data migrations or large-scale file transfers.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
