Structured troubleshooting
10GbE Performance Checklist: A Repeatable Troubleshooting Order
Random tuning creates random results. A repeatable checklist prevents you from replacing switches, disabling security or changing MTU before proving which layer is actually below target.
Quick answer
Start by isolating the layer, not by buying hardware
Random tuning creates random results. A repeatable checklist prevents you from replacing switches, disabling security or changing MTU before proving which layer is actually below target.
Current Amazon listings
Multi-gig hardware matched by exact model
Technical decision
The troubleshooting decision
Work from physical link to transport, then host bus, storage and protocol. Record each result and stop changing the network once network-only throughput is healthy.
Interactive troubleshooting tool
10GbE Performance Checklist
Enter measured values where possible. The result is a diagnostic priority, not a substitute for vendor support, cable certification, security policy or controlled testing.
Multi-gig checklist
Four checks before changing or replacing anything
Find the slowest link
End-to-end throughput cannot exceed the slowest NIC, switch port, uplink, cable path, host bus or storage endpoint in the flow.
Separate line rate from file speed
10GbE is 10 gigabits per second at the Ethernet layer, not a promise that every SMB copy will show 1.25 GB/s.
Test the network without storage
Use iPerf3 or an equivalent network-only benchmark before concluding that SMB, RAID or a disk is the reason the Ethernet path is slow.
Measure storage as well as network
A single HDD, small RAID, SATA SSD, NVMe pool and RAM cache can produce very different ceilings even on the same 10GbE link.
Record the target and current symptom
Write down the expected workload, current file speed, direction and whether the slowdown is constant or intermittent. A vague goal like “make 10GbE fast” is hard to troubleshoot.
For 10GbE Performance Checklist: A Repeatable Troubleshooting Order, checkpoint 1 records record the target and current symptom. Keep 10GbE Performance Checklist: A Repeatable Troubleshooting Order conditions fixed. Change one variable inside 10GbE Performance Checklist: A Repeatable Troubleshooting Order, then remeasure record the target and current symptom. If record the target and current symptom moves, 10GbE Performance Checklist: A Repeatable Troubleshooting Order stays on this layer. If record the target and current symptom remains stable, 10GbE Performance Checklist: A Repeatable Troubleshooting Order advances. That 10GbE Performance Checklist: A Repeatable Troubleshooting Order discipline protects 10GbE Performance Checklist: A Repeatable Troubleshooting Order from unrelated hardware changes.
Verify every negotiated Ethernet rate
Check client, switch, uplink and server/NAS interfaces. Correct product specifications do not matter if the live link trained at 1GbE or 5GbE.
Inspect physical-layer counters
Look for CRC errors, drops, link flaps and transceiver warnings. A clean negotiated rate with dirty counters is not a healthy baseline.
For 10GbE Performance Checklist: A Repeatable Troubleshooting Order, checkpoint 3 records inspect physical-layer counters. Keep 10GbE Performance Checklist: A Repeatable Troubleshooting Order conditions fixed. Change one variable inside 10GbE Performance Checklist: A Repeatable Troubleshooting Order, then remeasure inspect physical-layer counters. If inspect physical-layer counters moves, 10GbE Performance Checklist: A Repeatable Troubleshooting Order stays on this layer. If inspect physical-layer counters remains stable, 10GbE Performance Checklist: A Repeatable Troubleshooting Order advances. That 10GbE Performance Checklist: A Repeatable Troubleshooting Order discipline protects 10GbE Performance Checklist: A Repeatable Troubleshooting Order from unrelated hardware changes.
Run iPerf3 in both directions
Use a network-only test to decide whether the raw TCP path can approach the expected rate. Keep the command and duration consistent for future retests.
Check PCIe or external host bandwidth
Confirm a PCIe NIC is operating at the expected generation and width, or that an external adapter is connected through the required USB4/Thunderbolt path.
Benchmark source read and destination write
Measure local storage independently. The slower storage endpoint is a hard ceiling on file-copy throughput.
For 10GbE Performance Checklist: A Repeatable Troubleshooting Order, checkpoint 6 records benchmark source read and destination write. Keep 10GbE Performance Checklist: A Repeatable Troubleshooting Order conditions fixed. Change one variable inside 10GbE Performance Checklist: A Repeatable Troubleshooting Order, then remeasure benchmark source read and destination write. If benchmark source read and destination write moves, 10GbE Performance Checklist: A Repeatable Troubleshooting Order stays on this layer. If benchmark source read and destination write remains stable, 10GbE Performance Checklist: A Repeatable Troubleshooting Order advances. That 10GbE Performance Checklist: A Repeatable Troubleshooting Order discipline protects 10GbE Performance Checklist: A Repeatable Troubleshooting Order from unrelated hardware changes.
Review CPU and interrupt load
Watch both endpoints during iPerf3 and file tests. High CPU, encryption or virtual switching can cap throughput even when the wire is healthy.
Compare large sequential and small-file workloads
Use a large file for throughput baseline, then test the real workload. A small-file penalty is not automatically a network defect.
Inspect SMB or NFS features
Signing, encryption, Multichannel, RSS and mount/share settings can affect performance. Preserve security requirements and change only documented settings.
For 10GbE Performance Checklist: A Repeatable Troubleshooting Order, checkpoint 9 records inspect smb or nfs features. Keep 10GbE Performance Checklist: A Repeatable Troubleshooting Order conditions fixed. Change one variable inside 10GbE Performance Checklist: A Repeatable Troubleshooting Order, then remeasure inspect smb or nfs features. If inspect smb or nfs features moves, 10GbE Performance Checklist: A Repeatable Troubleshooting Order stays on this layer. If inspect smb or nfs features remains stable, 10GbE Performance Checklist: A Repeatable Troubleshooting Order advances. That 10GbE Performance Checklist: A Repeatable Troubleshooting Order discipline protects 10GbE Performance Checklist: A Repeatable Troubleshooting Order from unrelated hardware changes.
Test MTU only after the baseline is stable
Jumbo frames are an optimization, not a prerequisite. Confirm the complete path supports the larger MTU before enabling it.
Check scheduled NAS and server work
Scrubs, backups, indexing and antivirus can create time-dependent slowdowns. Reproduce tests outside maintenance windows.
Retest and document the final state
After a fix, repeat link, iPerf3, storage and application tests. Record the working driver, firmware, MTU, protocol and topology so future regressions are easier to diagnose.
For 10GbE Performance Checklist: A Repeatable Troubleshooting Order, checkpoint 12 records retest and document the final state. Keep 10GbE Performance Checklist: A Repeatable Troubleshooting Order conditions fixed. Change one variable inside 10GbE Performance Checklist: A Repeatable Troubleshooting Order, then remeasure retest and document the final state. If retest and document the final state moves, 10GbE Performance Checklist: A Repeatable Troubleshooting Order stays on this layer. If retest and document the final state remains stable, 10GbE Performance Checklist: A Repeatable Troubleshooting Order advances. That 10GbE Performance Checklist: A Repeatable Troubleshooting Order discipline protects 10GbE Performance Checklist: A Repeatable Troubleshooting Order from unrelated hardware changes.
Questions people ask
Common questions about 10GbE checklist
What should I check first on slow 10GbE?
Start with the actual negotiated link rate on every segment.
When should I run iPerf3?
Immediately after link and error checks, before spending time on storage or SMB tuning.
What if iPerf3 is healthy?
Stop treating the basic network path as the primary suspect and move to storage, protocol and CPU.
Should I enable MTU 9000 early?
No. Keep MTU 1500 for the initial stable baseline.
Should I disable SMB signing to test?
Do not disable required security by default. Measure and follow platform/security policy.
Can PCIe be checked after storage?
It can, but checking host-bus link state early is quick and prevents misleading NIC conclusions.
Why test both read and write directions?
NIC, storage and protocol behavior can be asymmetric, so one direction does not prove the other.
Why test outside backup windows?
Background I/O can dominate storage performance and produce false troubleshooting conclusions.
When do I replace hardware?
After the checklist identifies a specific component that remains below target across controlled tests.
What should I document after the fix?
Topology, negotiated speeds, driver/firmware versions, MTU, iPerf3, storage benchmarks and the final application speed.
Primary references and methodology
Verify the exact port, cable, host and storage path
Work from physical link to transport, then host bus, storage and protocol. Record each result and stop changing the network once network-only throughput is healthy. In 10GbE Performance Checklist: A Repeatable Troubleshooting Order, record the target and current symptom is measured before check pcie or external host bandwidth. 10GbE Performance Checklist: A Repeatable Troubleshooting Order then checks inspect smb or nfs features before retest and document the final state. The 10GbE Performance Checklist keeps 10GbE Performance Checklist: A Repeatable Troubleshooting Order measurements ordered. In 10GbE Performance Checklist: A Repeatable Troubleshooting Order, a passing layer is left untouched. A replacement purchase enters 10GbE Performance Checklist: A Repeatable Troubleshooting Order only after the measured diagnosis points there.
- QNAP: troubleshoot dramatic NAS transfer slowdowns
- ESnet iperf3 documentation and releases
- Microsoft: slow SMB file transfer troubleshooting
As an Amazon Associate, Cloudzat may earn from qualifying purchases. Ethernet line rate is not the same as application throughput. Product revisions, chipsets, firmware, operating-system drivers, cable quality, thermals, switch configuration and storage performance can change results; verify the exact hardware revision before purchase.