Since a servlet is blocking, a normal HTTP communication makes the thread wait. So there is a possibility that all the threads in the pool can be used up. By using the asynchronous http client framework Asynchronous HTTP Client framework (AHC javadocs), the thread can be released back into the pool and be made available for other purposes until the answer comes back. When the answer comes back, a callback is fired which then relays the response back to the client.

...

The AsyncServlet in this app uses the Asynchronous HTTP Client( AHC ) and implements the CometProcessor interface. The CometProcessor will make it's event method invoked rather than the usual service method, according to the event which occurred. The event object gives access to the usual request and response objects, which may be used in the usual way. More information about this can be read here.

...

for the module org.apache.geronimo.configs/tomcat6/<version>/car, add the following gbean

Code Block

	XML
	XML
borderStyle	solid

<gbean gbeanInfo="org.apache.geronimo.tomcat.connector.Http11NIOConnectorGBean"
       name="org.apache.geronimo.configs/tomcat6/2.0.1/car?ServiceModule=org.apache.geronimo.configs/tomcat6/2.0.1/car,j2eeType=GBean,name=TomcatNIOConnector">
<attribute name="useExecutor">true</attribute>
<attribute name="maxThreads">40</attribute>
<attribute name="pollerThreadCount">1</attribute>
<attribute name="socket_soTimeout">5000</attribute>
<attribute name="selectorTimeout">1000</attribute>
<attribute name="connectionTimeout">60000</attribute>
<attribute name="socket_soLingerOn">true</attribute>
<attribute name="acceptorThreadCount">1</attribute>
<attribute name="port">8080</attribute>
<attribute name="pollerThreadPriority">5</attribute>
<attribute name="compressableMimeType">text/html,text/xml,text/plain</attribute>
<attribute name="socket_performanceLatency">0</attribute>
<attribute name="useComet">true</attribute>
<attribute name="useIPVHosts">false</attribute>
<attribute name="maxHttpHeaderSize">4096</attribute>
<attribute name="restrictedUserAgents" value=""/>
<attribute name="acceptorThreadPriority">5</attribute>
<attribute name="proxyPort">0</attribute>
<attribute name="maxPostSize">2097152</attribute>
<attribute name="minSpareThreads">10</attribute>
<attribute name="socketBuffer">9000</attribute>
<attribute name="redirectPort">8443</attribute>
<attribute name="bufferSize">2048</attribute>
<attribute name="socket_soReuseAddress">false</attribute>
<attribute name="useSendfile">true</attribute>
<attribute name="threadPriority">5</attribute>
<attribute name="socket_tcpNoDelay">false</attribute>
<attribute name="useBodyEncodingForURI">false</attribute>
<attribute name="maxSpareThreads">100</attribute>
<attribute name="enableLookups">true</attribute>
<attribute name="socket_performanceConnectionTime">1</attribute>
<attribute name="socket_appWriteBufSize">8192</attribute>
<attribute name="maxKeepAliveRequests">100</attribute>
<attribute name="socket_ooBInline">true</attribute>
<attribute name="socket_keyCache">500</attribute>
<attribute name="disableUploadTimeout">true</attribute>
<attribute name="socket_bufferPoolSize">104857600</attribute>
<attribute name="keepAliveTimeout">60000</attribute>
<attribute name="acceptCount">100</attribute>
<attribute name="socket_txBufSize">43800</attribute>
<attribute name="connectionLinger">-1</attribute>
<attribute name="command_line_options">true</attribute>
<attribute name="tcpNoDelay">true</attribute>
<attribute name="oomParachute">1048576</attribute>
<attribute name="socket_appReadBufSize">8192</attribute>
<attribute name="selectorPool_maxSpareSelectors">-1</attribute>
<attribute name="xpoweredBy">false</attribute>
<attribute name="socket_rxBufSize">25188</attribute>
<attribute name="maxSavePostSize">4096</attribute>
<attribute name="socket_soLingerTime">25</attribute>
<attribute name="compression">off</attribute>
<attribute name="host">0.0.0.0</attribute>
<attribute name="socket_eventCache">500</attribute>
<attribute name="socket_performanceBandwidth">1</attribute>
<attribute name="noCompressionUserAgents" value=""/>
<attribute name="name">TomcatNIOConnector</attribute>
<attribute name="socket_soKeepAlive">false</attribute>
<attribute name="socket_soTrafficClass">28</attribute>
<attribute name="socket_directBuffer">false</attribute>
<attribute name="selectorPool_maxSelectors">200</attribute>
<attribute name="emptySessionPath">false</attribute>
<attribute name="uriEncoding">ISO-8859-1</attribute>
<attribute name="allowTrace">false</attribute>
<attribute name="processCache">200</attribute>
<attribute name="socket_processorCache">500</attribute>
<attribute name="server" null="true"/>
<attribute name="proxyName" null="true"/>
<reference name="TomcatContainer">
<pattern>
<groupId>org.apache.geronimo.configs</groupId>
<artifactId>tomcat6</artifactId>
<version>2.0.1</version>
<type>car</type>
<name>TomcatWebContainer</name>
</pattern>
</reference>
<reference name="ServerInfo">
<pattern>
<groupId>org.apache.geronimo.configs</groupId>
<artifactId>j2ee-system</artifactId>
<version>2.0.1</version>
<type>car</type>
<name>ServerInfo</name>
</pattern>
</reference>
</gbean>

The gbean uses version 2.0.1. Replace that with the appropriate version of your server
Start the server.

...

The following test results show a comparison between an AHC enhanced servlet and a normal blocking servlet. Each of these servlets are pounded by 16, 200 and 500 simultaneous requests. The output captured from the python script shows the response time for each request. These tests also demonstrate the significance of tuning the IP stacks. When none of the machines are tuned, the response time starts increasing as the number of threads start increasing. Executing a netstat command shows that most sockets are either in TIME_WAIT or CLOSED_WAIT state.