Design – Build – Run: Applied Practices and Principles for Production-Ready Software Development

8.4. Summary

In this chapter you looked at monitoring and the sources of monitoring information. You also learned which events and performance counters can be monitored, and received some food for thought regarding an application's custom performance counters.

The following are the key points to take away from this chapter:

• Monitoring is critical to live service. All production systems should be monitored effectively and the operations team is primarily responsible for all monitoring.

• Alerts are used to draw attention to the system. Alerts can be raised by the application itself or raised based on monitoring configuration and rules.

• Events can be positive and negative. You should think about both positive and negative events that might be required by operators.

• Reduce "noise." Monitoring rules should be implemented to reduce unnecessary alerts and noise. In addition, the application should ideally support monitoring configuration for the severity of its events. This allows tuning during testing as well as live service.

• Implement custom events. The application should include custom events. The high-level categories of events discussed included:

  • Informational

  • Warning

  • Error

  • Success and failure audit

• Implement custom performance counters. The application should include custom performance counters to improve monitoring in live service. The following types of performance counters were discussed:

  • Counters that are updated on a per-transaction basis

  • Counters that are updated on a per-connection ...