-sd-animation: sd-fadeIn; –sd-duration: 0ms; –sd-easing: ease-in;

Ethernet Throughput Utility: Troubleshoot Slow Network Speeds

Slow network speeds can stem from many causes: faulty cables, misconfigured devices, congested switches, or software bottlenecks. An Ethernet throughput utility helps you measure raw network performance between two endpoints and pinpoint where the slowdown occurs. This article explains how to use such a utility to diagnose and fix slow Ethernet speeds, step by step.

What an Ethernet throughput utility measures

  • Throughput (bandwidth): Actual data transfer rate (usually Mbps or Gbps).
  • Jitter: Variation in packet delay—important for real-time apps.
  • Packet loss: Percentage of packets that never arrive.
  • Latency (optional): Round-trip time between endpoints.

Tools you can use

Common utilities include iperf/iperf3, netperf, and built-in vendor tools. iperf3 is free, cross-platform, and widely used—this article assumes iperf3 but principles apply to other tools.

Preparation: setup and baseline checks

  1. Choose two endpoints: One acts as server, the other as client. Use wired connections for both.
  2. Minimize background traffic: Disable backups, updates, streaming, and large transfers during tests.
  3. Use proper cables and ports: Prefer at least Cat5e for 1 Gbps and Cat6/Cat6a for 10 Gbps. Verify ports support the intended speed.
  4. Update NIC drivers and firmware: Outdated drivers can limit throughput.
  5. Check link speed: On each endpoint, confirm the link negotiation (e.g., 100 Mbps, 1 Gbps, 10 Gbps). This sets an upper bound.

Running tests with iperf3 (example)

  1. On the server machine, run:
iperf3 -s
  1. On the client machine, run a basic test:
iperf3 -c 
  1. For a bidirectional test:
iperf3 -c  -d
  1. For parallel streams (to saturate multi-core systems or test TCP window limits):
iperf3 -c  -P 8
  1. For UDP testing (to measure packet loss/jitter):
iperf3 -c  -u -b 1G

Interpreting results

  • Throughput near negotiated link speed: Likely not a physical-layer issue. Look at switch congestion, server CPU limits, or application-level bottlenecks.
  • Throughput far below link speed with correct link negotiation: Check duplex mismatch, driver/firmware issues, or NIC offload settings.
  • High packet loss or jitter: Problems with cabling, electromagnetic interference, overloaded switches, or faulty NICs.
  • Asymmetric results (client→server different than server→client): Could indicate NIC configuration differences, duplex issues, or QoS rules.

Common causes and fixes

  • Bad or wrong cable: Replace with certified cable (Cat5e/6/6a) and retest.
  • Duplex or link-negotiation mismatch: Force correct speed/duplex only if auto-negotiation fails; usually better to keep auto.
  • Faulty ports or switch hardware: Test by moving endpoints to different switch ports or using a direct cable between machines.
  • Switch congestion or inadequate capacity: Inspect switch CPU, enable QoS, segment traffic, or upgrade switch.
  • CPU or NIC offload limitations: Monitor CPU during tests; disable certain offloads (TCP segmentation offload, large receive offload) if they cause issues.
  • Driver/firmware

Your email address will not be published. Required fields are marked *