<target name="benchmark" description="Runs the benchmarks" depends="tasks">
        <delete file="${benchmark.log}"/>
        <property name="threads" value="5"/>
        <property name="duration" value="60"/>
        <property name="delay" value="0"/>
        <taskdef name="benchmark" classname="net.sf.jbenchmark.BenchmarkTask">
            <classpath>
                <path refid="project.class.path"/>
            </classpath>
        </taskdef>

        <benchmark>
            <script path="scripts/benchmark.all.xml"/>
            <httpserver hostname="${benchmark.host}" port="${benchmark.port}" threadSession="true"/>
            <publisher logfile="${benchmark.log}"/>
        </benchmark>
    </target>

<benchmark> Ant Task

The <benchmark> task takes the following attributes:

Nested Element

The <benchmark> task takes the following nested elements:

<script>

The <script> element defines a benchmark script to execute. It takes the following attributes:

• Uses Jdk Logging.
• Logging can vastly affect performance metrics, typical use should be with a default level of SEVERE.

The benchmark tool is a simple HTTP benchmarking tool. Benchmarks are defined in an XML based benchmark script, and an Ant task is used to execute the script.

Sample 1 - Basic Page hits

<test>
    <!-- Hit the web page with multiple threads, for a fixed duration -->
    <web-page prefetchCount="15" prefetchDelay="500" threads="${threads}" duration="${duration}" delay="${delay}"
        path="/path/" name="index.xml">
        <assert-response-status status="200"/>
        <assert-response-size max="20000"/>
    </web-page>
</test>

Sample 2 - simulate workload.

<test xmlns:jelly="jelly:core" xmlns:util="jelly:util">
    <thread-group threads="5">
        <loop-duration duration="120">
            <jelly:import uri="include/sitesearch.xml" inherit="true"/>
        </loop-duration>
        <timer name="Preload times only (get the server and caches warmed up)"/>
    </thread-group>

    <thread-group threads="${threads}">
        <loop-duration duration="${duration}">
            <jelly:import uri="include/sitesearch.xml" inherit="true"/>
        </loop-duration>
        <timer name="search times"/>
    </thread-group>

file:include/sitesearch.xml

            <choose-random>
                <!-- 80% of hits are searches with partial property name-->
                <choice weight="80">
                    <!-- 50% of searches have accomodation type set -->
                    <random var="accommType">
                        <token-list tokens=",,,1,2,3"/>
                    </random>
                    <random var="propertyName">
                        <sql-list query="select substring(DisplayName, 1, 4) from PROPERY"/>
                    </random>
                    <http-get path="/Search.jsp?propertyName={propertyName}&accomodationType={accommType}">
                        <assert-response-status status="200"/>
                        <assert-response-size min="1000" max="600000"/>
                    </http-get>
                    <timer name="searchby.propname"/>
                </choice>
                <!-- 20% of searches by region -->
                <choice weight="20">
                    <random var="regionid">
                        <sql-list query="select id from REGION"/>
                    </random>
                    <http-get path="/Search.jsp?region={regionid}">
                        <assert-response-status status="200"/>
                        <assert-response-size min="1000" max="600000"/>
                    </http-get>
                    <timer name="searchby.region"/>
                </choice>
            </choose-random>

A benchmark script is an XML file. The root element of the script must be a <test> test group element.