Thanks RajDevies, I also used the default value (32mb) as you suggested, still this didn't help me. Do you think there are some other properties to combine with the syncOnWrite to make performance acceptable??
Thank you. rajdavies wrote: > > Although syncOnWrite will be slow - I think the size of the data logs > - 1000mb is very large. Its is possible that syncOnWrite will provide > acceptable performance for smaller data log size - have you tried it > with the default size ? > > > On Dec 21, 2007, at 4:18 AM, EricMeena wrote: > >> >> I was to a point where the AMQ performance was really impressive, >> but when I >> added some AMQ message storage configurations, the performance becomes >> questionable, I think I am combining wrong properties: >> >> I have two JMS clients(producer & receiver), I am sending about 10,000 >> messages (500KB each) and receiving them on the other side. >> This is what I am using: >> - MS Windows XP Pro, Version 2002, SP2, Core(TM)2 CPU, T 7200 @ >> 2.00GHz, >> 1.66 GHz, 1.99 GB of RAM >> - ActiveMQ v5 >> - Sun JVM 5 >> - My JMS clients use a spring JmsTemplate and a listenerContainer. >> (For >> connectionFactory, I am using this: >> org.apache.activemq.pool.PooledConnectionFactory). I am using a queue. >> - I am configuring the message storage within the default AMQ broker >> configuration file (activemq.xml) as follows: >> >> <persistenceAdapter> >> <amqPersistenceAdapter directory="C:/ActiveMQ/ActiveMQ_Storage" >> maxFileLength="1000mb" >> directoryArchive="C:/ActiveMQ/ActiveMQ_Storage_Archive" >> archiveDataLogs="true" syncOnWrite="true"/> >> </persistenceAdapter> >> >> Problem: >> >> Everything was fine until I added the sycOnWrite="true", I could >> send and >> consume up 17 messages per second. With this prop added, the sending >> itself >> becomes a little slow, but reading gets worse. The sender finishes >> its job >> while the consumer is not even half way done. This led to a >> performance of 1 >> message per second. >> I added few properties to fix this (like persistentIndex and >> indexBinSize), >> none helped the situation. >> >> Any idea? What is wrong here? >> >> >> Thanks in advance. >> -- >> View this message in context: >> http://www.nabble.com/ActiveMQ-v5.0.0-Performance-issue-tp14450031s2354p14450031.html >> Sent from the ActiveMQ - User mailing list archive at Nabble.com. >> > > > -- View this message in context: http://www.nabble.com/ActiveMQ-v5.0.0-Performance-issue-tp14450031s2354p14458455.html Sent from the ActiveMQ - User mailing list archive at Nabble.com.
