10.8. Workload Distribution

Distributing work optimally between a client and its server is an important challenge when architecting RIA. In traditional web applications, this decision was simple, as servers did most of the heavy lifting and clients were thin. With RIA things are no more as straightforward. Rich clients allow you to do a lot more than thin clients by maintaining local state and providing a rich interactive infrastructure. Greater interactivity often implies fewer trips back and forth between the client and the server.

Besides, you need to consider the fact that client machines today are more powerful than their counterparts a few years back. Current generation client machines have substantially larger CPU and memory resources at their disposal.

What then constitutes the right workload balance between the client and the server? While there is no universally correct answer to this question, this section attempts to lay out the criteria that govern and affect the decision of distributing work optimally between a rich client and its server.

Simply put, creating thin clients should not be preferred. Flex applications are capable of maintaining state locally and carrying out client-side interaction logic and responding to user gestures, especially by dispatching appropriate events.

Servers for Flex applications should serve the purpose of connecting to enterprise server-side resources, databases and other middle-tier workflow and rules engines. RESTful interfaces are preferred for loose coupling and scalability.