Got it running better by setting the split operation not parallel and by threading on the split results.
Here is the output from VisualVM: http://i.imgur.com/m4jFQdK.png Here is the output from Camel: http://i.imgur.com/1BZWl4b.png. The test starts at 10:55. The previous one is from the other test, with no .threads(). As you can see, it is much faster but it still drops. I don't know if a larger result set won't stop it because that slope tends to zero... I think the problem is that I'm creating a whole lot of objects just to create the XML. Maybe I should create the XML directly by string concatenation without creating the objects first. Maven Ran camel:run with four threads. BTW, I changed from SpringDSL to JavaDSL. @Component public class ReadDB extends SpringRouteBuilder{ @Override public void configure() throws Exception { from("quartz2://myGroup/myTimerName?trigger.repeatInterval=20000&trigger.repeatCount=0") .to("sql:select * from sales where integration_id = (select max(integration_id) as max_int from sales)") .split(simple("${body}")).streaming() .to("myLogger") .to("direct:sqlqueue"); } } public class WriteToQueue extends SpringRouteBuilder{ @Override public void configure() throws Exception { from("direct:sqlqueue") .threads(4) .to("lineProcessor") .marshal("xstream-utf8") .to("rabbitmq://xx.xx.xx.xx:5672/myExchange?username=xxxx&password=xxxx") ; } } -- View this message in context: http://camel.465427.n5.nabble.com/Camel-gradually-slowing-down-with-large-sql-results-tp5741560p5741594.html Sent from the Camel - Users mailing list archive at Nabble.com.
