Best iPerf3 Alternatives for Network Testing

Before we pick a winner, what’s the event?

If you have ever searched for a professional network testing tool, there is a good chance you ended up with iPerf3, which is not surprising. It’s been around for a long time and yes, it’s free, powerful, and many network professionals know enough of its command-line syntax to make it do whatever they need.

But "widely used" and "perfect for every job" are two very different things. Perhaps you want official Windows support, live charts instead of rows of numbers, or built-in historical reporting. So, what else is out there?

Let’s Choose the Candidates

Don’t we all love it when some software or hardware vendor publishes an “unbiased” “Top 10 tools” list comparing their product to a bunch of carefully selected competitors? Naturally, when you reach the summary table you find out that 9 out of 10 tools suffer from multiple deficiencies, and only one of the products (the vendor’s own product) is really good and can beat all the others with one hand tied behind its back. Well, maybe it has just one tiny drawback, like “hard to find, sells so well that it’s often out of stock” or “available in 12 colors, but adding pink would be nice”. Familiar? Okay, that’s not what I’m going to do here.

Here, I applied the sort of rigorous scientific methodology, ahem, ahem, you've come to expect from this blog. I went to Google and searched for “best network throughput testing software”. No, seriously, I can’t think of a more unbiased method.  If you know one – let me know. Meanwhile, here is what Google gave me (sorry about the huge screenshot):

Our highly sophisticated market-research department at work

Granted, your Google output may be different from mine, and the list will evolve in the future (as it evolved in the past; just a week ago the same question gave me, among others, a very old and unmaintained product, LAN Speed Test, but today it’s no longer present).

That said, I still want to add just one tool, Keysight IxChariot, because otherwise we would be comparing products from roughly the same end of the market and ignoring what large-scale enterprise network testing looks like. With this addition, I’ve summarized the data for a quick comparison:

Tool	Interface	TCP	UDP	Main strength	Main limitation
iPerf3	CLI	Yes	Yes	Powerful, free, scriptable testing	No official Windows support; CLI workflow
Tessabyte	GUI + CLI	Yes	Yes	Continuous testing, reports, QoS, cross-platform workflow	Commercial client
OpenSpeedTest	Browser	Browser-based	No traditional UDP test	Extremely easy client-side deployment	Less control over traffic generation
Netperf	CLI	Yes	Yes	Throughput plus request/response tests	Aging tool and documentation
Keysight IxChariot	GUI/Web	Yes	Yes	Large-scale application traffic emulation	Enterprise complexity and pricing

Now let's compare the tools, see where each tool actually fits and also look at what Google got right and wrong.

1. iPerf3: The Benchmark Every Other Tool Has to Measure Up Against

Let's start with the elephant in the server room, so to speak. iPerf3 is good software: It performs active measurements of achievable network bandwidth, supports TCP and UDP, parallel streams, reverse and bidirectional tests, socket and buffer tuning, JSON output, and lots of other options. Plus, it’s by no means an abandoned utility hanging around because someone forgot to uninstall it in 2009.

The classic iPerf3 workflow is wonderfully simple once you know it. Start a server on one machine, run the client on another, and tell it what kind of traffic to generate.

iperf3 -c 192.168.1.10 -u -b 500M

Perfectly clear, if you already know that -c means client, -u means UDP, and -b means target bitrate. Want reverse traffic? Add -R. Want several parallel streams? Add -P. Want DSCP? Another parameter.

Will you remember what all of this means six months down the road? No? Well, then welcome to the brief archaeological expedition through the documentation. I don’t intend to criticize command-line software. It clearly has merits, but sometimes you might need something else.

Google Says Windows, ESnet Says Something Else

Anyway, now we arrive at the first interesting part of Google's answer. Google's screenshot lists iPerf3 platforms as: Windows, Linux, macOS, Android, FreeBSD. But there is a small problem with that: The official iPerf3 documentation says that its supported platforms are Ubuntu Linux, FreeBSD, and macOS. It specifically states that iPerf3 is not officially supported on Windows. Community Windows binaries do exist, and people certainly use them, but "somebody compiled it for Windows" and "the project officially supports Windows" are not the same thing. So Google is not completely hallucinating; you can run iPerf3 on Windows and Android, but its platform list quietly ignores an important distinction between "it can run there" and "the developers support it there."

Another good example is DSCP/QoS testing. On officially supported operating systems, iPerf3 provides DSCP-related options. On Windows, however, QoS marking has long been one of the problematic areas. The iPerf3 issue tracker contains a specific report about DSCP/QoS tagging not working on Windows, and this is exactly the kind of situation where "Windows is unsupported" stops being a footnote and becomes relevant. We covered this in more detail in our earlier Moving Your Network Tests from iPerf3 to Tessabyte article.

So, does that make iPerf3 bad? Of course not. If you're working primarily on Linux or macOS, like the command line, and want a highly configurable tool that can easily be scripted, iPerf3 remains an excellent choice.

2. Tessabyte: Less Typing, More Testing

Before we continue, the disclosure: Tessabyte is our product. There. Nobody can accuse us of secretly being Netmantics. We didn't insert Tessabyte into the list ourselves, Google did. It even correctly identified it as "Tessabyte Throughput Test (formerly TamoSoft Throughput Test)." That part is right, but its platform information is not. Google lists Tessabyte simply as: Windows.

Well... no. Tessabyte Client runs on Windows, macOS, and iOS/iPadOS. Tessabyte Server runs on Windows, macOS, and Linux, and there is a multi-architecture Docker server image as well.

So, Google managed to make iPerf3 look more cross-platform than its official support actually is while simultaneously making Tessabyte look less cross-platform than it really is. AI symmetry, perhaps.

What Is Different From iPerf3?

At the lowest level, the idea is familiar. You put a server at one end of a network path and a client at the other, then generate traffic between them. The major difference is the workflow. Tessabyte continuously measures TCP and/or UDP traffic in both directions and displays current and average throughput, UDP packet loss, RTT, and jitter on live charts.

To be clear, iPerf3 can also run continuously and report measurements at regular intervals. Tessabyte’s advantage here is that you get the charts, stored measurements, and reporting tools together, without having to assemble that workflow yourself.

That makes a difference when you're actually troubleshooting something. Imagine you're testing Wi-Fi while walking around an office. With Tessabyte, leave the test running and walk. When throughput falls, RTT rises, or jitter suddenly goes crazy behind a particular wall, you can see the change on the chart and compare it with what was happening a moment earlier. Sometimes a dip in a line tells the story faster than twenty rows of numbers.

We use the same principle for wired troubleshooting. If you're trying to find a network bottleneck, move the endpoints, change one part of the path at a time, and watch how the measurements respond. We recently wrote an entire article about that workflow.

TCP, UDP, QoS and Reports

Tessabyte can test TCP only, UDP only, or both simultaneously. You can rate-limit traffic, select QoS/DSCP traffic classes, control test parameters, and connect multiple clients to the same server. UDP testing includes packet loss and RFC 3550 jitter measurements.

And unlike Google's description, it isn't just about looking at a pretty chart: All measurements are stored in a database. Windows and macOS versions can generate HTML or CSV reports covering a selected time range and destination, while the iOS/iPadOS version can export PDF or CSV reports. That's an important distinction if the test is part of your job rather than a five-minute curiosity.

If a customer says, "The Wi-Fi was terrible yesterday afternoon," seeing 742 Mbps right now doesn't help much, but a historical report might.

The disadvantage is obvious: Tessabyte is commercial software. If all you need is to run a quick TCP test between two Linux servers and print Mbps in a terminal, it is difficult to beat iPerf3's price of exactly zero dollars. But if you want a proper Windows/macOS client, continuous bidirectional testing, charts, UDP metrics, QoS testing, scheduling, and reports, then the comparison begins to look rather different.

3. OpenSpeedTest: Because Sometimes You Just Want a Browser

Google's third suggestion is OpenSpeedTest, and this one approaches the problem from a completely different angle. With OpenSpeedTest, you can host the test server yourself and then run the client side from a web browser. There are server applications for Windows, macOS, Linux, iOS, and Android, as well as Docker and source-code options for headless deployment. The project says the self-hosted version can operate entirely inside a LAN, without relying on an external Internet speed-test server.

And this solves a very real problem: Suppose you're testing Wi-Fi in an office and someone hands you a laptop, two phones, an iPad, a Chromebook, and a mysterious handheld warehouse terminal designed in 2014 by a company that no longer exists. Installing special test software on all of those devices could become your afternoon, while opening a browser and typing an address takes just five seconds, and that's exactly OpenSpeedTest's superpower!

The self-hosted version measures upload and download performance against your own server, and its documentation says it has been tested at up to 10 Gbps on physical NICs and 30 Gbps on virtual NICs.

But all nice things, including cigars and browser convenience, come with a trade-off. What’s the trade-off here? Well, OpenSpeedTest is fundamentally a browser-based speed test and it’s good at answering:

"How fast can this device communicate with this server?"

It is not designed to give you the same low-level control over UDP load, packet-loss behavior, DSCP/QoS classes, protocol combinations, or other test parameters that you'd expect from iPerf3 or Tessabyte. So, if you want to test many types of client devices with virtually zero client-side setup, OpenSpeedTest may be exactly the right compromise. But if your question is:

"At what UDP rate does loss begin, what happens to jitter, and does Voice-marked traffic survive the path?"

then you’re probably knocking on the wrong door.

4. Netperf: The Older Relative Who Still Remembers Some Clever Tricks

Google's fourth suggestion is Netperf. It’s another classic network benchmark, originally developed at Hewlett-Packard. It can measure unidirectional bulk throughput, but it also provides request/response tests designed to examine end-to-end transaction latency. That is an important difference from a simple "fill the pipe and tell me the Mbps" test.

Think of it this way.

A throughput test asks:

"How quickly can I move a truckload of boxes from A to B?"

A request/response test asks:

"How quickly can I send one box to B, receive an acknowledgement, and repeat that process thousands of times?"

Those are not the same performance questions. This makes Netperf useful for network and system tuning where transaction rate and latency matter as much as raw throughput. There is, however, a certain... vintage quality to the project.

The Netperf repository itself repeatedly warns readers that the manual may be outdated. Its release notes identify the current 2.7.0 release line, and the project documentation contains plenty of material whose age is clearly visible. That doesn't mean the underlying benchmark has suddenly stopped measuring TCP correctly. Good low-level networking tools can remain useful for a very long time. After all, nobody refuses to use ping because its user interface hasn't been redesigned with rounded corners😊

But Netperf feels like an engineering benchmark, not a modern network-testing application. The user experience is command-line driven, the documentation is not especially welcoming to newcomers, and deployment requires more effort than the newer options in this article. If you're a Linux/Unix engineer interested in throughput and request/response benchmarking, Netperf still deserves a place in the toolbox.

If your requirement is "give this to the technician in the branch office and ask him to test the Wi-Fi," it would not be my first thought.

5. Keysight IxChariot: When a Bigger Hammer Is Actually Appropriate

Google didn't give us this one. I added it myself. Why? Because the first four tools are all, in one way or another, tools you might use to test communication between a relatively small number of endpoints. That leaves out an entirely different class of network testing.

Enter Keysight IxChariot.

Calling IxChariot an "iPerf3 alternative" is technically true but slightly misleading. It is like calling an airliner an alternative to a bicycle because both can move you from one place to another.

IxChariot is designed to emulate real application traffic across distributed software endpoints. It can model flows such as HTTP, VoIP, video, and file transfers, and measure throughput, latency, jitter, packet loss, application latency, and quality-of-experience metrics. Keysight currently offers configurations from a single endpoint pair all the way to 500 pairs.

That lets you ask much larger questions.

Instead of:

"Can this link sustain 1 Gbps?"

you can investigate something closer to:

"What happens when hundreds of endpoints generate voice, video, web, and data traffic across our branches and cloud infrastructure simultaneously?"

Keysight positions IxChariot specifically for this kind of realistic, multi-endpoint traffic emulation across wired, wireless, cloud, branch, and enterprise environments. That is an entirely different class of testing.

It also comes with an entirely different class of deployment, licensing, complexity, and price. Keysight's current product pages use quote-based purchasing for IxChariot configurations rather than the sort of price tag you expect from a desktop utility.

If you're validating a Wi-Fi 7 rollout across a campus or modelling hundreds of application flows across an SD-WAN deployment, suddenly the giant hammer begins to make sense.

So Which iPerf3 Alternative Should You Use?

After all of this, you may be disappointed to learn that there is no universal answer.

If you're happy with the command line, primarily use Linux/macOS/FreeBSD, want something free and scriptable, and don't need a larger reporting workflow, iPerf3 may still be exactly the tool you should use.

If the client device has a browser and you simply want a quick local upload/download measurement without installing client software, OpenSpeedTest is hard to beat for convenience.

If you're interested in request/response performance as well as bulk throughput and you're comfortable with old-school CLI networking tools, Netperf still offers useful capabilities.

If you need hundreds of distributed endpoints and realistic application traffic emulation, IxChariot is in a different class from the other tools here.

And if your job is testing LANs, Wi-Fi, WAN links, VPNs, routers, firewalls, or other network paths and you want TCP and UDP testing, live charts, packet loss, jitter, RTT, QoS, reports, scheduling, proper Windows/macOS support, and a workflow you don't need to reconstruct from command-line history, that is exactly the problem Tessabyte was built to solve.

The Real Question Isn't "Which Tool Is Best?"

Network engineers love benchmarks, including benchmarks of benchmarking tools, but asking which one of these products is "best" without specifying the job is almost meaningless.

A hammer is better than a screwdriver when there is a nail in front of you. This does not establish the hammer's general superiority over screwdrivers.

The more useful questions are what you are trying to measure, whether you control both endpoints, whether you need TCP or UDP, whether loss and jitter matter, whether you need QoS testing, whether Windows is involved, whether you need to keep the results, and whether you're testing two endpoints or two hundred.

Once those questions are answered, the choice usually becomes much easier.

And when it comes to network performance, there is an even better option:

---

Test it yourself.

If you want to see what Tessabyte can do on your own LAN, Wi-Fi, WAN, or VPN, download Tessabyte and run the same tests yourself. The Windows and macOS desktop versions include a 30-day evaluation period, so you don't have to take our word for any of this.

---