Benchmarking Network Throughput

Ever wonder if your internet connection is truly delivering what it promises? Or if your local network is a bottleneck?

Network benchmarking helps you measure and understand your network's actual performance. It's crucial for diagnosing slowdowns, verifying upgrades, or planning system deployments.

When we talk about network performance, we often focus on these key metrics:

• Throughput (Bandwidth): The amount of data transferred successfully per unit of time, usually measured in bits per second (bps) or bytes per second (Bps).
• Latency: The time it takes for a data packet to travel from one point to another.
• Jitter: The variation in latency over time.
• Packet Loss: The percentage of data packets that fail to reach their destination.

For benchmarking, throughput is often the primary focus.

iperf3 is a powerful command-line tool specifically designed for measuring maximum network throughput between two hosts.

It's widely used because it can test both TCP and UDP connections, show detailed statistics, and is available on almost all platforms, including Linux.

To test network throughput with iperf3, you need two machines: one acting as a server and one as a client.

First, on one of your Linux machines (the 'server'), start iperf3 in server mode. It will listen for incoming connections.

Now, on your second Linux machine (the 'client'), run iperf3 in client mode, pointing it to the server's IP address.

This command will perform a default 10-second TCP test, sending data from the client to the server.

After the test, both the client and server will display results. Look for the 'Bandwidth' column, which shows the throughput.

Key metrics in the output:

• Interval: Time segments of the test.
• Transfer: Total data sent/received in that interval.
• Bandwidth: The average data rate for that interval.
• Retr: Number of retransmitted packets (for TCP), indicating network issues.

iperf3 can also test UDP performance, which is useful for applications like video streaming or VoIP where packet loss is more critical than retransmissions.

Use the -u flag on the client to switch to UDP mode. You can also specify a target bandwidth with -b.

Sometimes, a single TCP stream might not fully utilize your network's capacity due to TCP windowing or latency.

You can use the -P flag to run multiple parallel streams simultaneously, which can reveal the true maximum throughput of your link.

iperf3 offers many other options for fine-tuning your tests:

• -R: Reverse mode (server sends, client receives).
• -t : Set test duration.
• -i : Set interval for reporting.
• -J: Output results in JSON format for scripting.

Always check iperf3 --help for a full list of options!

You've run an iperf3 TCP test from a client to a server. The client output shows a 'Retr' count of 150. What does this indicate?

In this lesson, we explored network throughput benchmarking. We learned:

• Why benchmarking is essential for understanding network performance.
• Key metrics like throughput, latency, jitter, and packet loss.
• How to use iperf3 in both server and client modes.
• Interpreting iperf3 results, including bandwidth and retransmissions.
• Testing UDP performance and using parallel streams for maximum throughput.

Mastering iperf3 is a vital skill for any network professional or developer!