May 2018 Summaries

Harness has significantly enhanced its integration with AppDynamics to provide comprehensive service impact verification across all microservices in an application environment, allowing businesses to better understand the implications of each deployment. This integration enables organizations to instantly assess the upstream and downstream effects of microservices deployments, improving deployment reliability and reducing potential downtime. AppDynamics offers automatic mapping of application service dependencies through tracing user requests, which Harness leverages to evaluate microservice dependencies and their performance impacts. With this enhanced capability, Harness employs unsupervised machine learning to analyze time-series metrics and assess the real-time performance of microservices, providing a full view of the deployment's impact. This integration allows for automatic rollback of deployments when verifications indicate a negative impact, thus preventing downtime and ensuring a seamless user experience.

The OODA loop framework, initially developed by Col. John Boyd for military strategy, emphasizes the importance of rapid decision-making through the stages of Observe, Orient, Decide, and Act, and has been adapted for business environments to enhance product development. Companies like Netflix, Facebook, and LinkedIn exemplify this approach by deploying changes frequently, yet the focus is often skewed towards the "Act" phase, potentially leading to rapid but ineffective solutions. To address this, it is crucial for product and engineering teams to define metrics that measure the long-term value of their products, ensuring that each step in the development process is aligned with creating value for the end user. Experimentation, as practiced by companies like Microsoft and Netflix, is highlighted as a key method for completing the OODA loop by allowing teams to gather insights and iterate effectively. Additionally, tools like the Split Feature Data Platform™ support this process by enabling feature management and experimentation, ensuring changes are beneficial and safe, thereby revolutionizing product delivery without sacrificing quality.

Integrating Harness with Datadog allows for automated verification of cloud application and infrastructure performance across deployment pipelines, significantly reducing manual verification time from 60 minutes to 15 minutes and enabling automatic rollbacks in production deployments. Harness aids in mastering Continuous Delivery by enabling swift and confident deployment of new applications and services, while Datadog provides monitoring of cloud application performance. The integration facilitates automatic verification of performance across various deployment stages, from development to production, highlighting any performance regressions and allowing for quick remedial actions. A practical example of this integration shows a deployment pipeline where a new microservice version is verified at each stage; if a performance issue is detected, as was the case with a key web transaction regression, an automatic rollback is triggered. Harness employs unsupervised machine learning techniques to identify performance deviations, and the integration is easily configurable via the Datadog API. Future enhancements are anticipated to include additional features such as deployment markers for Datadog users and support for unstructured event data, further enhancing the capabilities of this integration.