Hello, You are saying that the HTTP traffic is consumed quickly. Maybe the bottleneck is the ActiveMQ broker itself. Try to disable broker's persistence as described in: http://activemq.apache.org/persistence.html .
Kind Regards, Marek Psiuk Gert Vanthienen wrote: > > Ben, > > Could you check the thread pool sizes with jconsole? Perhaps one of > the pools just isn't large enough and is slowing everything down. > Perhaps it's easier to start with one defaultConfig with some very > high defaults (like core=16, queue=0, max=128) and then start > fine-tuning things by adding more specific and strict configurations > for some components. Also, if you're not using the JMS/JCA flow, > perhaps you could disable those to ease the configuration effort. > > Regards, > > Gert Vanthienen > ------------------------ > Open Source SOA: http://fusesource.com > Blog: http://gertvanthienen.blogspot.com/ > > > > 2009/7/31 boday <[email protected]>: >> >> here is an update on this... >> >> By using a different HTTP endpoint (servicemix-http instead of Camel's) >> I'm >> able to get better multi-threaded execution and slightly higher CPU >> utilization on this Solaris box (~10-15%). The issue is that my overall >> throughput is still extremely low. Any ideas about what could cause >> this? >> Are there any known issues with Solaris? >> >> I'm able to process ~150-200 msg/sec on my laptop (2.4Ghz, Core 2 Duo, >> 4GB), >> but am only seeing roughly ~60 msg/sec with the identical build on this >> Solaris box. Supposedly this box should be able to handle a >> substantially >> higher level of traffic (8 CPU core's, 8GB RAM, etc.). Have tried both >> Sun >> and JRockit JVMs as well...no difference. >> >> I have been playing around with the threading configuration (increases >> 3-4x >> what I have setup on my machine). Still the CPU load stays low on this >> machine (10% vs. 80% on my laptop) and throughput is terrible. It is >> able >> to consume the HTTP traffic quickly, but just queues up in front of my >> JBI >> service that processes the data. >> >> The HTTP component is the only one able to really get high throughput, >> everything else just queues up downstream. I have increased my thread >> configuration as follows... >> >> <property name="configs"> >> <map> >> <entry key="flow.jms.servicemix-http"> >> <bean >> class="org.apache.servicemix.executors.impl.ExecutorConfig"> >> <property name="corePoolSize" value="2"/> >> <property name="maximumPoolSize" value="4"/> >> <property name="queueSize" value="1024"/> >> </bean> >> </entry> >> <entry key="flow.jms.servicemix-jms"> >> <bean >> class="org.apache.servicemix.executors.impl.ExecutorConfig"> >> <property name="corePoolSize" value="32"/> >> <property name="maximumPoolSize" value="64"/> >> <property name="queueSize" value="1024"/> >> </bean> >> </entry> >> <entry key="flow.seda.servicemix-bean"> >> <bean >> class="org.apache.servicemix.executors.impl.ExecutorConfig"> >> <property name="corePoolSize" value="64"/> >> <property name="maximumPoolSize" value="128"/> >> <property name="queueSize" value="1024"/> >> </bean> >> </entry> >> <entry key="flow.seda.servicemix-camel"> >> <bean >> class="org.apache.servicemix.executors.impl.ExecutorConfig"> >> <property name="corePoolSize" value="32"/> >> <property name="maximumPoolSize" value="64"/> >> <property name="queueSize" value="1024"/> >> </bean> >> </entry> >> </map> >> </property> >> >> thanks in advance... >> >> ----- >> Ben - Senior Consultant >> >> -- >> View this message in context: >> http://www.nabble.com/ServiceMix-multi-threaded-execution-tp24508532p24759936.html >> Sent from the ServiceMix - User mailing list archive at Nabble.com. >> >> > > > ----- > --- > Gert Vanthienen > http://gertvanthienen.blogspot.com > -- View this message in context: http://www.nabble.com/ServiceMix-multi-threaded-execution-tp24508532p25764189.html Sent from the ServiceMix - User mailing list archive at Nabble.com.
