The Role of Monitoring

Regardless of how well a network, computer system, or an application is configured and maintained, hardware and software elements can quit functioning or perform poorly. Situations such as server crashes, memory leaks, and disk failures are a reality in today's computing environments.

The main purposes of monitoring are to:

• Flag a failed server, or one nearing a pre-defined critical threshold.
• Flag a successful (or failed) operation.
• Provide a tool for identifying performance bottlenecks.
• Eliminate the guesswork in capacity planning by providing tangible metrics.

Member and cluster-wide monitoring capabilities are essential to managing complex and dynamic systems that support mission-critical applications. Tools for monitoring different components in a production environment have been available for several years now, and although they can be adapted to handle server clusters, the overhead involved in juggling multiple tools and sets of data grows exponentially with the size and complexity of the clusters.

The guiding principles for the Application Center monitoring feature are:

• The ability to instrument a member based on tangible and measurable events.
• Provide metrics based on real-time data from any cluster member.
• Enable automated actions, such as e-mail notifications and smart recovery.
• Provide flexible access to real-time and historical data.

In Chapter 6, "Synchronization and Deployment," you saw how Application Center provides real-time and results-based information for the synchronization service and application deployment. Other chapters—in particular, Chapter 10, "Working with Performance Counters"—deal with the performance aspect of cluster monitoring. This chapter focuses on the tools that you can use to monitor computer and cluster health.