Portal Server and High Availability
System Availability
System availability is often expressed as a percentage of the system uptime. A basic equation to calculate system availability is:
Availability = uptime / (uptime + downtime) * 100
For instance, a service level agreement uptime of four digits (99.99 percent) means that in a month the system can be unavailable for about seven hours. Furthermore, system downtime is the total time the system is not available for use. This total includes not only unplanned downtime, such as hardware failures and network outages, but also planned downtime, preventive maintenance, software upgrade, and patches.
If the system is supposed to be available seven days a week, 24 hours a day, the architecture needs to include redundancy to avoid planned and unplanned downtime to ensure high availability.
Degrees of High Availability
High availability is not just a switch that you can turn on and off. Various degrees of high availability refer to the ability of the system to recover from failures and ways of measuring system availability. The degree of high availability depends on your specific organization’s fault tolerance requirements and ways of measuring system availability.
For example, your organization might tolerate the need to reauthenticate after a system failure, so that a request resulting in a redirection to another login screen would be considered successful. For other organizations, this might be considered a failure, even though the service is still being provided by the system.
Session failover alone is not the ultimate answer to transparent failover, because the context of a particular portal application can be lost after a failover. For example, consider the case where a user is composing a message in NetMail Lite, has attached several documents to the email, then the server fails. The user is redirected to another server and NetMail Lite will have lost the user’s session and the draft message. Other providers, which store contextual data in the current JVM™, have the same problem.
Achieving High Availability for Portal Server
Making Portal Server highly available involves ensuring high availability on each of the following components:
Chapter 5 Creating Your Portal Design 85
