Summary of best practices

Summary of best practices

• In designing the application server tier, consider keeping the application server tier stateless, that is, no state management by the application server tier . This means the state could be handled on the client or in a back end database. Advantages of stateless over stateful are:

  • Best horizontal scalability

  • Higher availability in maintenance and failover scenarios

  • Overall ease of system management, for example adding and remove servers to a cluster

• Run multiple instances of an application per node. There are two possible benefits:

  • If one instance fails, the hardware capacity is not totally lost

  • For large SMP machines (four CPUs or higher), higher transaction throughput may be realized

• Deploy critical applications in their own application server instance and, for small machines, keep applications on physically separate nodes. This configuration provides higher application reliability.

• For performance reasons, do not allow application servers to serve static content. However, to keep maintenance simple, deploy static content with the application EARs and use edge side caching (ESI) features either in the Web server, in Caching Proxy Edge component, or an external caching service provider to serve the content. This is appropriate when there is strong interdependence between the static and dynamic content for each application release or upgrade level. Maintenance is simple because static content is deployed in the same step as the application, keeping the static and dynamic content association.

• Create scripts to automate as many processes as possible to eliminate human error during critical procedures.

• To improve response time, serve static and dynamic content from the same Web servers when the content makes up the same Web page. This makes full use of keep alive by ensuring that the browser does not have to connect to a different cluster for static and dynamic content.

• Ensure that in a cluster, Web, or application, consisting of two physical nodes that each node is capable of handling the site's peak load in the event of a failure of one of the nodes.

• For better scalability and performance, configure load balancers for special purpose. For example, multiple tiers of load balancers can be used to improve scalability. Tier one does content-based routing to multiple pools of servers while multiple load balancers at tier two perform round robin to each individual pool of servers.

Similar products