Outils de performance réseau

Ever wonder how fast your network truly is, or what's happening behind the scenes? In this lesson, we'll dive into powerful Linux tools to measure and understand network performance.

We'll explore iperf for bandwidth, netstat for connections, and ss for detailed socket statistics.

Network performance isn't just about speed! It includes several key metrics:

• Throughput: The amount of data transferred over time (e.g., Mbps).
• Latency: The delay before a transfer of data begins (e.g., ms).
• Jitter: Variation in latency, which can affect real-time applications.
• Packet Loss: Data packets failing to reach their destination.

Measuring these helps diagnose slow or unreliable connections.

iperf is a fantastic tool for measuring network throughput. It works in a client-server model, where one machine acts as a server and another as a client, sending data between them.

This allows you to test the raw speed and capacity of a link between two points in your network.

To use iperf, you first need to start a server on one machine. It will listen for incoming connections from clients.

Open a terminal on your server machine and run this command:

Now, on a different machine (your client), you can connect to the iperf server and start a bandwidth test. Replace <server_ip> with the actual IP address of your server.

The output will show you the transfer rate and bandwidth achieved.

iperf isn't just for TCP! You can also test UDP throughput and observe packet loss or jitter. UDP is often used for real-time applications where speed is critical.

To run a UDP test, add the -u flag. Try it out!

netstat (network statistics) is a classic command-line tool that displays network connections, routing tables, interface statistics, and more.

It's great for quickly seeing what ports are open and what connections are active on your system.

To see all active TCP and UDP connections, along with listening ports, use the following flags:

• -t: TCP connections
• -u: UDP connections
• -l: Listening sockets
• -n: Numeric addresses (don't resolve hostnames)

This gives a raw, fast overview.

While netstat is useful, ss (socket statistics) is its modern, faster, and more powerful successor. It can display more information about sockets than netstat.

ss directly queries kernel data structures, making it quicker and more efficient, especially on systems with many connections.

ss uses similar flags to netstat but often provides more detail or performs faster. Here are a couple of common uses:

• ss -tuln: Same as netstat, shows TCP/UDP listening ports.
• ss -s: Provides a summary of socket statistics (e.g., total TCP established, UDP in use).

Give ss -s a try to see a quick summary!

You've learned about iperf, netstat, and ss. Which of these tools is primarily used to measure network bandwidth or throughput?

Great job! You've explored essential tools for network performance:

• iperf: Your go-to for measuring network throughput (bandwidth) and testing latency/jitter with TCP and UDP.
• netstat: A classic for viewing active connections, listening ports, and basic network statistics.
• ss: The modern, faster alternative to netstat for detailed socket information.

These tools are invaluable for diagnosing network issues and ensuring optimal performance.