JMeter vs Gatling Tool

I think you'll agree when I say:

It's REALLY hard to decide whenever JMeter or Gatling Tool should be used.

You're probably wondering:

• How does JMeter to Gatling Tool compare?
• Which tool has best documentation? Performances? Script Maintainability?
• Should I use JMeter or Gatling? Or maybe both?

Well, it turns out you can gain significant insight on JMeter and Gatling differences in just 5 minutes reading!

Following up our article about why we chose JMeter to build OctoPerf, our Cloud Load Testing Platform, this post compares JMeter and Gatling Tool on many different fields:

• Project History: which tools is improved and maintained more actively,
• Features: which of the two has the most features,
• Setup: how each tool can be installed,
• Documentation: which tool does have the most complete documentation,
• Scripting: compares how to design scripts with both tool,
• Performances: is JMeter faster than Gatling? Which is the most efficient tool,
• and even more!

Ready for some action? Let's go!

OctoPerf is JMeter on steroids!

Schedule a Demo

Project History

This section covers historical data about these two open-source projects.

JMeter History

JMeter has been first released in 1998, so this project more than 15 years old. It's a mature project which has released more than 20 versions. JMeter is purely written in Java, although OpenHub shows that it also contains XML and HTML:

JMeter has more than 10.000 commits representing almost 500K lines of code. 33 people contributed to the project so far. Philippe M. is one of the biggest contributor to the project, mostly in Java. He is the founder of Ubik Load Pack, a French company offering commercial JMeter plugins.

JMeter is open-source and released under the Apache 2.0 license.

Gatling Tool History

Gatling Tool has been first released on December 20, 2011. Gatling is a younger project, with almost 4 years old codebase. Gatling is mostly written in Scala.

Scala differs for Java by offering full support for Functional programming. Scala has been created by a Martin Odersky, as German computer scientist.

Stéphane Landelle is the main contributor and creator of Gatling tool. More than 60 people contributed to this project. Its size is almost 50K lines of code.

Gatling tool is open-source and released under the Apache 2.0 license. Its development is supported by Excilys.

History Winner

JMeter wins on Project History. The project is significantly older and has more contributors. Both Gatling and JMeter are maintained regularly. JMeter has a significantly bigger codebase (almost 500K lines now), which Gatling tool is much smaller (50K lines of code). Of course, Scala code is more concise than Java code.

Features

JMeter Features

JMeter has obviously a great support for many different protocols:

• Ability to load and performance test many different applications/server/protocol types:

• Web: HTTP, HTTPS (Java, NodeJS, PHP, ASP.NET, …),

• SOAP / REST Webservices,
• FTP,
• Database via JDBC,
• LDAP,
• Message-oriented middleware (MOM) via JMS,
• Mail - SMTP(S), POP3(S) and IMAP(S),
• Native commands or shell scripts
• TCP,
• Full featured Test IDE that allows fast Test Plan recording (from Browsers or native applications), building and debugging.
• Command-line mode (Non GUI / headless mode) to load test from any Java compatible OS (Linux, Windows, Mac OSX, …)
• Dynamic HTML report,
• Easy correlation through ability to extract data from most popular response formats, HTML, JSON , XML or any textual format,
• Complete portability and 100% Java purity,
• Full multi-threading framework allows concurrent sampling by many threads and simultaneous sampling of different functions by separate thread groups,
• Caching and offline analysis/replaying of test results.

Let's review Gatling Tool Features now.

Gatling Tool Features

Gatling has the following interesting features:

• Standalone HTTP Proxy Recorder,
• Scala-based scripting,
• An expressive self-explanatory DSL for test development,
• asynchronous non-blocking engine for maximum performance,
• Excellent support of HTTP(S) protocols and can also be used for JDBC and JMS load testing,
• Validations and assertions,
• and a Comprehensive HTML Report.

You probably wonder now: How does JMeter Features compare to Gatling Features?

Features Comparison Table

The following Table compares the features supported by both tools.

Feature	JMeter	Gatling
HTTP	X	X
HTTPS	X	X
WebSocket	X	X
Rest Apis	X	X
Soap Services	X
FTP	X
JDBC	X
Database via JDBC	X
JMS	X	X
TCP	X
HTTP Recorder	X	X
HTML Reporting	X	X
Distributed Tests	X
GUI Scripting	X
Code Scripting		X
Command-Line Mode	X	X
Assertions	X	X
Response Extractors	X	X

X: external plugins may be required.

Features Verdict

And the winner is... JMeter! JMeter has clearly more features than Gatling Tool. This is completely understandable considering both project histories.

So sure, JMeter has more features. But, do you really need them all?

Let's see now how each tool performs when it comes to teaching how to use them.

Documentation

The match between JMeter and Gatling tool ends with a draw on documentation. Both are great! You may prefer the look of Gatling's documentation, but JMeter has recently made a great effort to refresh the look of their website.

JMeter Documentation

The JMeter documentation is really great. Every feature of the tool is detailed with every possible setting. They truly made a huge effort on explaining how JMeter works, how to create your first test and more.

Gatling Documentation

The Gatling Documentation is very clear and concise, like the philosophy of Scala. There is really nothing more to say, the documentation talks about any feature in depth and covers the whole tool.

Documentation Comparison Table

The following table compares JMeter and Gatling Tool documentations:

Documentation	JMeter	Gatling
Installation	X	X
Quick Start	X	X
Reference Guide	X	X
Tutorials	X	X
Issue Tracker	X	X
On Stackoverflow	X	X

As you see it, both tools have great documentation. JMeter and Gatling Tool are active on StackOverFlow, a question answering community.

Documentation Verdict

Both JMeter and Gatling Tool have great documentation. The result of this battle is... draw!

Setup

JMeter and Gatling tool both require a Java JRE installed on the machine.

JMeter Setup

• Download JMeter Archive,
• Unzip the archive,
• Run ./bin/jmeter.sh executable. (.bat on Windows)

Gatling Tool Setup

• Download Gatling Tool Archive,
• Unzip the archive,
• Run ./bin/gatling.sh executable. (.bat on Windows)

Setup Comparison Table

The following Table compares the setup process:

Setup	JMeter	Gatling
Archive Format	X	X
Installer Format
Command-line Executable	X	X
GUI Executable	X
Requires Java	X	X

Setup Verdict

As you can see, both tools have comparable setup. When it comes to Tool Setup, the result is draw!

Both are JVM applications that must be installed as a standalone program on your machine. There is no real difference in installing JMeter or Gatling. Both are shipped as archives, no additional installation is required once you have a working Java JRE.

Scripting

One big difference between JMeter and Gatling is the project format. Gatling saves them (called Simulation) as Scala classes with scala extension.

JMeter Scripting

Here is what a JMeter JMX looks like:

<?xml version="1.0" encoding="UTF-8"?>
<jmeterTestPlan version="1.2" properties="2.8" jmeter="2.13 r1665067">
  <hashTree>
    <TestPlan guiclass="TestPlanGui" testclass="TestPlan" testname="Test Plan" enabled="true">
      <stringProp name="TestPlan.comments"></stringProp>
      <boolProp name="TestPlan.functional_mode">false</boolProp>
      <boolProp name="TestPlan.serialize_threadgroups">false</boolProp>
      <elementProp name="TestPlan.user_defined_variables" elementType="Arguments" guiclass="ArgumentsPanel" testclass="Arguments" testname="User Defined Variables" enabled="true">
        <collectionProp name="Arguments.arguments"/>
      </elementProp>
      <stringProp name="TestPlan.user_define_classpath"></stringProp>
    </TestPlan>
    <hashTree>
      <ThreadGroup guiclass="ThreadGroupGui" testclass="ThreadGroup" testname="Thread Group" enabled="true">
        <stringProp name="ThreadGroup.on_sample_error">continue</stringProp>
        <elementProp name="ThreadGroup.main_controller" elementType="LoopController" guiclass="LoopControlPanel" testclass="LoopController" testname="Loop Controller" enabled="true">
          <boolProp name="LoopController.continue_forever">false</boolProp>
          <stringProp name="LoopController.loops">1</stringProp>
        </elementProp>
        <stringProp name="ThreadGroup.num_threads">1</stringProp>
        <stringProp name="ThreadGroup.ramp_time">1</stringProp>
        <longProp name="ThreadGroup.start_time">1433521552000</longProp>
        <longProp name="ThreadGroup.end_time">1433521552000</longProp>
        <boolProp name="ThreadGroup.scheduler">false</boolProp>
        <stringProp name="ThreadGroup.duration"></stringProp>
        <stringProp name="ThreadGroup.delay"></stringProp>
      </ThreadGroup>
      <hashTree>
        <HTTPSamplerProxy guiclass="HttpTestSampleGui" testclass="HTTPSamplerProxy" testname="HTTP Request" enabled="true">
          <elementProp name="HTTPsampler.Arguments" elementType="Arguments" guiclass="HTTPArgumentsPanel" testclass="Arguments" testname="User Defined Variables" enabled="true">
            <collectionProp name="Arguments.arguments"/>
          </elementProp>
          <stringProp name="HTTPSampler.domain">google.Com</stringProp>
          <stringProp name="HTTPSampler.port"></stringProp>
          <stringProp name="HTTPSampler.connect_timeout"></stringProp>
          <stringProp name="HTTPSampler.response_timeout"></stringProp>
          <stringProp name="HTTPSampler.protocol"></stringProp>
          <stringProp name="HTTPSampler.contentEncoding"></stringProp>
          <stringProp name="HTTPSampler.path"></stringProp>
          <stringProp name="HTTPSampler.method">GET</stringProp>
          <boolProp name="HTTPSampler.follow_redirects">true</boolProp>
          <boolProp name="HTTPSampler.auto_redirects">false</boolProp>
          <boolProp name="HTTPSampler.use_keepalive">true</boolProp>
          <boolProp name="HTTPSampler.DO_MULTIPART_POST">false</boolProp>
          <boolProp name="HTTPSampler.monitor">false</boolProp>
          <stringProp name="HTTPSampler.embedded_url_re"></stringProp>
        </HTTPSamplerProxy>
        <hashTree/>
      </hashTree>
    </hashTree>
  </hashTree>
</jmeterTestPlan>

JMeter saves projects as an XML file with jmx extension. See JMeter JMX Format for more information.

JMeter is GUI oriented. Opening, editing and saving a JMX file must be made with JMeter Graphical interface.

Do not attempt to modify the JMX yourself, even if the XML is human readable, it's really not sustainable. JMeter's GUI probably suits well for people accustomed to graphical tools.

Here is what JMeter graphical interface looks like:

JMeter JMX is more verbose than Gatling Scala simulation. But, you're not intended to read the JMX file directly. JMeter has a GUI to open those file, and display them. Gatling simulations are expected to be edited and modified directly.

Gatling Scripting

Here is what a Gatling simulation looks like:

package computerdatabase // 1

import io.gatling.core.Predef._ // 2
import io.gatling.http.Predef._
import scala.concurrent.duration._

class BasicSimulation extends Simulation { // 3

  val httpConf = http // 4
    .baseURL("http://computer-database.herokuapp.com") // 5
    .acceptHeader("text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8") // 6
    .doNotTrackHeader("1")
    .acceptLanguageHeader("en-US,en;q=0.5")
    .acceptEncodingHeader("gzip, deflate")
    .userAgentHeader("Mozilla/5.0 (Windows NT 5.1; rv:31.0) Gecko/20100101 Firefox/31.0")

  val scn = scenario("BasicSimulation") // 7
    .exec(http("request_1")  // 8
    .get("/")) // 9
    .pause(5) // 10

  setUp( // 11
    scn.inject(atOnceUsers(1)) // 12
  ).protocols(httpConf) // 13
}

In contrast to JMeter, Gatling Tool has no GUI. Editing the simulation requires you to write code. Gatling tool is clearly made for developers, although they claim it can be used by non programmers too. The Gatling simulation Scala syntax is pretty easy to understand, but as a developer so point of view is a little biased.

I would definitely recommend to use Scala IDE to edit Gatling Simulations. Intellij IDEA originally designed for Java also supports Scala via a plugin.

Running a Gatling simulation:

ubuntu@desktop:~$ ./bin/gatling.sh
Choose a simulation number:
     [0] computerdatabase.BasicSimulation
     [1] computerdatabase.advanced.AdvancedSimulationStep01
     [2] computerdatabase.advanced.AdvancedSimulationStep02
     [3] computerdatabase.advanced.AdvancedSimulationStep03
     [4] computerdatabase.advanced.AdvancedSimulationStep04
     [5] computerdatabase.advanced.AdvancedSimulationStep05

Simply hit 0 then Enter, and it runs the BasicSimulation script.

Scripting Comparison Table

The following table compares JMeter and Gatling on Scripting:

Scripting	JMeter	Gatling
Code	X
UI		X

JMeter is clearly oriented toward UI Scripting, while Gatling Tool is oriented toward Code Scripting. Each tool target a different audience:

• JMeter: mainly intended to be used by QA Engineers,
• Gatling Tool: mainly intended to be used by Devops Engineers.

Scripting Verdict

You've probably heard this a million times: the scripting winner depends on your feelings.

Do you prefer a point-and-click tools or Write Code? Depending on your preference, one tool may be better than the other.

Performance

JMeter Performance

JMeter uses the One Thread = One User Paradigm. It means that each simulated virtual user inside the load generator equals to a single thread. Wait times are simulated by Thread.sleep. The user waits for the server response when sending a request. This is sometimes called Synchronous processing.

Gatling Performance

Gatling uses a more advanced engine based on Akka. Akka is a distributed framework based on the actor model. It allows fully asynchronous computing. Actors are small entities communicating with other actors through messaging. It can simulate multiple virtual users with a single Thread. Gatling also makes use of Async HTTP Client.

Gatling is technically superior to JMeter. The programming model is definitely more innovant.

Performance Verdict

There was a debate about JMeter versus Gatling performance. As of today, none is performing better than the other.

Sure, the asynchronous thread model adopted by Gatling reduces CPU and memory usage a little bit, but usually the bottleneck comes from Network Bandwidth. Simulating 1K concurrent users on a single machine with 1GBps is equivalent to giving roughtly 1MBps per user. This is clearly much lower than DSL or 4G Bandwidth.

We decided to give the point to Gatling Tool. The innovative approach to the threading model, although bringing only minor benefits, merits attention.

Distributed Testing

You may ask: Which tool does support distributed testing?

Distributed Testing is the ability to use several slaves computers as load generators. The idea is to spread the load over multiple machines to be able to simulate more concurrent users.

JMeter Distributed Testing

When load testing an HTTP application, once you need to simulate several thousands users a single machine may not enough. The load must be splitted among multiple machines. JMeter and Gatling tool are clearly not behaving the same way on this field.

JMeter is distributed. JMeter uses a master / slave architecture. The master is connected to slaves via the RMI protocol. The master sends the JMX to simulate to each slave. Each slave plays the JMX. If you have 6 slaves, and your JMX simulates 1000 users, then you end up playing 6x1000 = 6000 users. Results are sent back to the master for consolidation.

If you want more info, see the Distributed Testing with JMeter tutorial.

Gatling Distributed Testing

Gatling is single host. Although Gatling is based on Akka which is a distributed actor framework, Gatling is not distributed. It cannot scale horizontally. You can of course launch multiple Gatling instances on different machines at the same time, but it's fully manual. You will also need to consolidate the results from the multiple injectors manually.

Distributed load testing is planned for Gatling, but only as a commercial feature.

Distributed Testing Verdict

JMeter stands out! Gatling Tool simply doesn't support distributed testing in the free version. Only Gatling Frontline, the paid Saas version, supports it.

HTTP Recorders

Both JMeter and Gatling Tool have built-in HTTP Proxy Recorders. The principle is quite simple:

The proxy is a man-in-the-middle technique to intercept both client requests and server responses. Those requests and responses are then stored as a script.

JMeter Recorder

JMeter has a built in proxy recorder to generate the requests from the user navigation.

If you want more information, see the JMeter Proxy step by step guide.

Gatling Recorder

Gatling has a separate Web proxy recorder, which works the same way as the JMeter recorder. It generates a Scala simulation from the recorded user session.

It has an interesting feature which allows to convert an HAR archive into a Gatling simulation.

Recorder Winner

The result of Recorder battle is draw. No tool clearly stands out on this point. Both have working HTTP recorders.

Test Execution

JMeter Test Execution

To launch a test with JMeter, there are two methods:

• Launch the GUI, then launch the test inside GUI,
• Launch the test via command-line (recommended).

JMeter offers various ways to view the results when using the GUI, for example with a graph:

But the recommended way is to launch JMeter in command-line, and save results as JTL files. Using GUI listeners consumes a lot of memory.

Since JMeter 3.0, HTML Reports are available too!

Those reports include many interesting performance metrics and charts like:

• A Statistics table providing in one table a summary of all metrics per transaction including 3 configurable percentiles,
• An error table providing a summary of all errors and their proportion in the total request,
• Response times Over Time (Includes Transaction Controller Sample Results) chart,
• and more.

The best way is to use a BackendListener which exports results to Grafana or InfluxDB:

Gatling Test Execution

To launch a test with gatling, run the .sh script and select the simulation (within user-files/simulations folder):

During the test run, gatling displays statistics within the console. The results of the test can be viewed at the end of the test. Gatling generates a generic test report with response time, number of OK/KO transactions and detailed statistics for every request. The report is not customizable.

Like JMeter, Gatling can be plugged into external analysis tools too!

Gatling supports Grafana with InfluxDB and Graphite along with their commercial offer named Frontline.

Test Execution Comparison Table

Let's sum up the Runtime comparison with a table:

Runtime	JMeter	Gatling
Command-line live metrics	X	X
HTML Report	X	X
Raw Results	X	X
UI Results	X
Graphite Support	X	X
Grafana + InfluxDB	X	X

Test Execution Verdict

Both tools are performing similarly when it comes to running tests. The only single advantage JMeter has is the ability to use UI Results Listeners.

The Runtime result is draw. Although JMeter has a little more results features,

Cloud Load Testing

JMeter Cloud

There are multiple companies that provide Cloud Load Testing for JMeter including brands like Blazemeter, Flood IO or RedLine 13.

All of them are just providing a facility to scale tests into the Cloud and provide a basic results reporting. You upload your JMX scenario or Gatling simulation to those providers, and they run your scenario on Cloud machines.

OctoPerf is world first JMeter Saas Company, providing JMeter Web UI Scripting. We also Scale from multiple locations and provide fully customizable test reports.

Gatling Tool Cloud

Some of the Cloud solutions exposed in the previous point also support Gatling Tool. Gatling Frontline is the defacto recommended standard for running Gatling tests in the cloud.

Cloud Load Testing Verdict

JMeter has better Cloud Support. Many solutions, even the expensive HP Storm Runner, are having support for JMEger tests.

Which Tool Should I Use?

Before answering the crucial question, let's review all the scores:

Winner On	JMeter	Gatling
History	X
Features	X
Documentation	X	X
Setup	X	X
Scripting	X	X
Performance		X
Distributed Testing	X
HTTP Recorders	X	X
Test Execution	X	X
Cloud Load Testing	X	X
Total	9 / 10	7 / 10

JMeter is clearly the winner. With greater number of Features and support for Distributed Testing, JMeter compensates for slightly less performances than Gatling Tool. JMeter's Longer Presence can be seen with two different point of views:

• The tool is old: True, JMeter is older and thus may be obsolete at some point,
• The tool is mature: Thanks to years of support and maintenance, the tool is stable and battle-tested.

Choosing JMeter or Gatling tool really depends on what you are comfortable with. JMeter is QA Engineers oriented while Gatling is Devops Engineers oriented.

Both tools are overall really great. The end choice really comes to your own feeling. Which Tool Am I comfortable with? The first answer coming out of your mind is probably the good one.

Want to become a super load tester?

Request a Demo