A recurring theme I hear from IT specialists that work with WebSphere Application Server administrators is the ongoing need to focus on quality of service. Avoiding application outages is critical to their customer satisfaction. However, achieving that goal can present real challenges to administrators, who must make contingency plans for unexpected outages as well as scheduled outages. Fortunately a talented team of authors has written an IBM Redpaper with information on addressing these challenges.
If you want to improve application resiliency in WebSphere Application Server, consider the 5 tips described briefly here. You can find these techniques described in more detail in Improve Application Resiliency with WebSphere Network Deployment V8.5. I highly recommend that you take a look at the information in this paper and apply the techniques described to your own environment.
Run multiple versions of an application concurrently to manage the rollout of new code
The application edition management features allow you to run multiple editions of an application concurrently and selectively route different users to a specific edition. A validation mode gives you the ability to test the new edition in your production environment without exposing it to all of your users. After validation is complete, the new application edition can be rolled out into regular production mode without service interruption for your users. If things don't go so well, you have the ability to roll back to a previous edition with a single click
Prioritize business critical applications and absorb spikes in application load
The intelligent management features can help you protect the performance of business critical applications by prioritizing applications and dynamically scaling resources to handle unanticipated spikes in application load. You define application service levels to meet user-defined business goals, and using these service policies, WebSphere Application Server can queue requests for less important applications to allow requests for more important applications to be handled quicker. WebSphere Application Server dynamically manages workload by starting and stopping additional application server instances to accommodate changes in load and to balance processing power among your applications to best meet your defined performance goals
Correct common problems before outages occur
The health monitoring features can be used to sense and respond to common problems before user outages occur. Health monitoring can detect conditions that indicate a problem with an application server, such as memory leaks, hung threads, slow response times, or an excessive percentage of request time outs. Other conditions, such as age or work performed, can be monitored to prevent problems in a server hosting an application with known minor errors. The conditions to monitor and the actions to take when they are encountered are defined in customizable and configurable health policies.
Eliminate single points of failure for administrative functions
Redundant deployment managers can be defined in an active-standby configuration. In this topology, an on demand router (ODR) routes administrative requests to the active deployment manager. If the active deployment manager is stopped or fails, the loss is detected and the standby is dynamically switched to active mode so it can take over administrative responsibility. A shared workspace ensures that work is not lost, and the ODR sends new requests to the newly activated deployment manager.
Improve application resiliency of batch workloads
The existing OLTP infrastructure in WebSphere Application Server provides features you can leverage with batch workloads, such as reusing business services through co-location, container-based checkpoint and restart capability, and exploiting system-managed batch parallelization. Converting eligible WebSphere batch workloads to run as parallel sub-jobs can help improve elapsed time. Also, be aware that you can include WebSphere batch jobs in your enterprise workload scheduler plan by using the WSGrid utility.