July 2019 Summaries

Dynatrace Software Intelligence offers a modern, AI-driven approach to monitoring that contrasts with older, manual-intensive tools like Nagios and Solarwinds. Unlike these first-generation tools, which rely heavily on user-defined alerting rules and manual thresholds, Dynatrace leverages AI to automatically detect and baseline the normal behavior of applications and services, reducing unnecessary alerts and focusing on issues that truly impact end-users. This shift represents a significant change for companies accustomed to traditional monitoring, as Dynatrace's AI adapts to frequent issues and adjusts alerting to avoid spam and irrelevant notifications. By automating the detection and analysis of anomalies, Dynatrace helps streamline the monitoring process, although it primarily applies to services and applications rather than infrastructure metrics. The transition to this AI-based system can be challenging but ultimately enhances monitoring efficiency and effectiveness.

Dynatrace utilizes its OneAgent to automatically detect and monitor deployed applications by analyzing running processes and mapping them to process groups, which represent individual applications. The system can identify various microservices, from Spring Boot-based REST services to business processes in IBM's enterprise ESB, and integrates with ServiceNow for automated IT operations. However, service detection can lack context due to missing information about microservices, either because of unsupported technologies or insufficient data provided by development teams. To address these gaps, Dynatrace has introduced a service detection API, enabling users to supplement or correct information for improved detection. This API allows for more precise configuration of service detection, including separate services based on subdomains, URLs, unmonitored hosts, or context roots. Additionally, the new API replaces the previous service merging functionality for better support in dynamic environments and allows for easy export and import of detection rules across environments. Future enhancements will expand capabilities to web service rules, and users are encouraged to provide feedback for further improvements.

The author describes their experience using keptn, an open-source control plane for continuous deployment and automated operations, on a Google Kubernetes Engine (GKE) cluster, highlighting its integration with Knative and Dynatrace for enhanced monitoring and automated operations. The deployment faced issues when Dynatrace alerted high CPU usage, traced back to Prometheus.knative-monitoring pods. Initially overlooked as part of demo activities, the problem persisted, revealing a performance issue with Prometheus. Automated analysis by Dynatrace identified these pods as the root cause, prompting the author to delete non-essential deployments in the knative-monitoring namespace, significantly reducing CPU consumption. The author reflects on the need for optimization within keptn, particularly with Prometheus, and acknowledges Dynatrace's effective problem detection and root cause analysis. They conclude by sharing the actions taken to address the issue and express optimism for future improvements in keptn’s resource management and performance.

Dynatrace has introduced a feature called rage click detection to identify user frustration in web and mobile applications, which typically manifests when users repeatedly click on unresponsive UI controls. This behavior highlights negative user experiences that may not be evident through traditional performance and error analysis. While Dynatrace already provides thorough insights into application performance and operational health, the rage click detection, when used alongside Session Replay, allows for a deeper understanding of usability issues by pinpointing the exact causes of user frustration, such as misleading buttons or server errors. This feature leverages AI-backed service insights to trace issues through the entire stack, offering a unique approach to improving user experience. Users can filter rage click events in session details and access them via the API or web interface, enabling companies to identify and address usability problems more effectively. Early-access users have already benefited from detecting issues related to responsiveness and browser-specific errors, and Dynatrace plans to expand its Real User Monitoring capabilities with additional user behavior insights in future updates.

Dynatrace offers an advanced approach to failure detection in services, utilizing its artificial intelligence engine, Davis®, to automatically identify and report issues such as HTTP 5xx errors and uncaught exceptions, unlike traditional monitoring systems that rely on manual rules. This capability is particularly effective for web services where developers adhere to good coding practices. However, the system can be fine-tuned to handle real-world complexities by configuring settings to ignore non-critical exceptions or include important application-specific ones. It also allows the customization of error detection rules for systems that do not rely on HTTP, such as backend processing services, by using request attributes to flag issues based on return codes. Furthermore, Dynatrace provides flexibility in how HTTP error codes are managed, allowing users to define specific codes for different service interactions to ensure accurate failure detection. The platform emphasizes the importance of maintaining high-quality data to enhance the effectiveness of failure detection, offering a 15-day free trial to new users.