Verify the NAS port is really at 10Gbps

Check the NAS interface status, switch port and client NIC. If any segment negotiated lower, solve that link problem before interpreting RAID or SMB benchmarks.

For 10GbE NAS Slow: Test Network, Storage and SMB Separately, checkpoint 1 records verify the nas port is really at 10gbps. Keep 10GbE NAS Slow: Test Network, Storage and SMB Separately conditions fixed. Change one variable inside 10GbE NAS Slow: Test Network, Storage and SMB Separately, then remeasure verify the nas port is really at 10gbps. If verify the nas port is really at 10gbps moves, 10GbE NAS Slow: Test Network, Storage and SMB Separately stays on this layer. If verify the nas port is really at 10gbps remains stable, 10GbE NAS Slow: Test Network, Storage and SMB Separately advances. That 10GbE NAS Slow: Test Network, Storage and SMB Separately discipline protects 10GbE NAS Slow: Test Network, Storage and SMB Separately from unrelated hardware changes.

Use iPerf3 to remove disks from the first test

QNAP explicitly recommends network-only testing when NAS transfers drop. A healthy iPerf3 result indicates the physical and TCP network path is likely capable, even if the file service is not.

Benchmark NAS storage without the network

Use vendor tools or a controlled local benchmark to understand what the pool can actually read and write. A small HDD array may be the dominant limit even though the NAS has a 10GbE interface.

For 10GbE NAS Slow: Test Network, Storage and SMB Separately, checkpoint 3 records benchmark nas storage without the network. Keep 10GbE NAS Slow: Test Network, Storage and SMB Separately conditions fixed. Change one variable inside 10GbE NAS Slow: Test Network, Storage and SMB Separately, then remeasure benchmark nas storage without the network. If benchmark nas storage without the network moves, 10GbE NAS Slow: Test Network, Storage and SMB Separately stays on this layer. If benchmark nas storage without the network remains stable, 10GbE NAS Slow: Test Network, Storage and SMB Separately advances. That 10GbE NAS Slow: Test Network, Storage and SMB Separately discipline protects 10GbE NAS Slow: Test Network, Storage and SMB Separately from unrelated hardware changes.

Benchmark the client drive as well

The workstation must read source data and accept destination data at comparable speed. Copying from a slow external drive to a fast NAS can make the NAS look slow when the sender is the bottleneck.

Check RAID state and drive health

A degraded array, rebuild, scrub or error-recovery workload can cut available I/O. Performance troubleshooting should pause until the storage pool is healthy enough that the benchmark represents normal operation.

Inspect scheduled jobs and indexing

Backups, snapshots, media indexing, deduplication, antivirus and cloud sync can compete with a foreground transfer. Correlate slow periods with the NAS task scheduler and resource monitor.

For 10GbE NAS Slow: Test Network, Storage and SMB Separately, checkpoint 6 records inspect scheduled jobs and indexing. Keep 10GbE NAS Slow: Test Network, Storage and SMB Separately conditions fixed. Change one variable inside 10GbE NAS Slow: Test Network, Storage and SMB Separately, then remeasure inspect scheduled jobs and indexing. If inspect scheduled jobs and indexing moves, 10GbE NAS Slow: Test Network, Storage and SMB Separately stays on this layer. If inspect scheduled jobs and indexing remains stable, 10GbE NAS Slow: Test Network, Storage and SMB Separately advances. That 10GbE NAS Slow: Test Network, Storage and SMB Separately discipline protects 10GbE NAS Slow: Test Network, Storage and SMB Separately from unrelated hardware changes.

Compare large files with small files

Large sequential files are the best first test of throughput. Small-file copies stress metadata, latency and filesystem operations, so a low MB/s number can be normal for that workload.

Measure SMB separately from raw TCP

If iPerf3 is fast but SMB is slow, examine SMB signing, encryption, Multichannel, CPU utilization and share configuration. Avoid changing security settings until you can demonstrate their effect.

Check memory pressure and cache behavior

NAS caching can make short benchmarks look faster than sustained work, while low memory and heavy services can create contention. Use long enough tests to expose the steady-state behavior.

For 10GbE NAS Slow: Test Network, Storage and SMB Separately, checkpoint 9 records check memory pressure and cache behavior. Keep 10GbE NAS Slow: Test Network, Storage and SMB Separately conditions fixed. Change one variable inside 10GbE NAS Slow: Test Network, Storage and SMB Separately, then remeasure check memory pressure and cache behavior. If check memory pressure and cache behavior moves, 10GbE NAS Slow: Test Network, Storage and SMB Separately stays on this layer. If check memory pressure and cache behavior remains stable, 10GbE NAS Slow: Test Network, Storage and SMB Separately advances. That 10GbE NAS Slow: Test Network, Storage and SMB Separately discipline protects 10GbE NAS Slow: Test Network, Storage and SMB Separately from unrelated hardware changes.

Confirm expansion cards and PCIe lanes

If the NAS uses an add-in 10GbE NIC, verify that the supported slot, card and negotiated PCIe link meet the platform requirements. A compatible connector is not proof of adequate host bandwidth.

Consider multi-client aggregate throughput

A NAS can justify 10GbE even when one HDD-backed client never reaches 1GB/s. Several 2.5GbE clients may collectively use the 10GbE uplink more effectively than one sequential stream.

Upgrade the layer that the measurements identify

If iPerf3 is slow, focus on network hardware. If local disk tests are slow, focus on drives/RAID. If both are fast but SMB is slow, focus on protocol and CPU behavior.

For 10GbE NAS Slow: Test Network, Storage and SMB Separately, checkpoint 12 records upgrade the layer that the measurements identify. Keep 10GbE NAS Slow: Test Network, Storage and SMB Separately conditions fixed. Change one variable inside 10GbE NAS Slow: Test Network, Storage and SMB Separately, then remeasure upgrade the layer that the measurements identify. If upgrade the layer that the measurements identify moves, 10GbE NAS Slow: Test Network, Storage and SMB Separately stays on this layer. If upgrade the layer that the measurements identify remains stable, 10GbE NAS Slow: Test Network, Storage and SMB Separately advances. That 10GbE NAS Slow: Test Network, Storage and SMB Separately discipline protects 10GbE NAS Slow: Test Network, Storage and SMB Separately from unrelated hardware changes.