Scalability of J2EE web services, how to approach?

I have this web client (built entirely through a Javascript Ajax framework), that communicates with the middle tier via web service calls (3 web services are deployed on JBoss, 1 web service is deployed on Tomcat). All of the web services access the same Oracle database. The web services are built through Axis2 and S开发者_运维知识库pring. All the services are deployed on 1 machine, the database on another.

I started running some SoapUI load tests, and noticed that the system starts dropping many connections once it reaches beyond the 200 concurrent users mark. What scalability approaches should I be researching in order to optimize the system to handle much larger volume of user? I know scalability is a very broad issue but, could you please provide me with a list of key areas I should start looking into and learning about?

Thanks!

Load balance across multiple application servers. There are several ways to do this, depending on how much you care about multi-site multi-server reliability and how much about pure performance scaling.