Modern applications deployments have increased in complexity due to distributed architectures with microservices. Increased number of microservices with multiple developers involved in development and release makes it difficult to track the root cause of failures. Metrics may not always reflect the underlying issue in the initial test, and root cause analysis requires distributed log analysis.
The OpsMx Autopilot performs automated log and metrics analysis for new releases with built-in unsupervised and supervised machine learning algorithms for risk analysis.
Autopilot is a release verification platform that provides Dev/Ops engineers an intelligent automated real-time actionable risk assessment of a new release deployed. The Autopilot verifies latest version of the service comparing to the baseline or prior release after production rollout. The baseline can be deployment from a prior time or current production instance during rollout using canary or blue/green or rolling update strategies.
It leverages unsupervised and supervised machine learning and Artificial Intelligence (AI) techniques to analyze 1000’s of metrics (infra and APM) and logs data to perform in-depth analysis of architectural regressions, performance, scalability and security violations of new releases in a scalable way for enterprises.
Autopilot exposes REST interface for triggering analysis, status queries and retrieving reports. These interfaces provide an easy to integrate with deployment pipelines from Spinnaker, Jenkins and other delivery tools. Automated analysis is triggered using the Analysis interface (CAS interface) and reports are accessed using Reporting service.
Autopilot provides interfaces for retrieving reports and query for status of runs in progress. The GUI provides easy to navigate and access capabilities for viewing reports and analyzing logs for diagnostics.
Autopilot Admin interface allows creating Data Source accounts for fetching monitoring information, creating Application groups and permissions and default templates. Users can also modify templates and applications based on permissions assigned.
AutoPilot Continuous Verification (CV) is a REST service that can be deployed on premise or use managed cloud service for analysis. AutoPilot CV interfaces with monitoring systems for logs and metrics and uses the metadata provided in start analysis phase to retrieve the logs and metrics for deployment verification. AutoPilot CV does not interface with the services deployed directly for its analysis.
AutoPilot CV can be deployed on premise as a Docker container that can run in Docker Swarm, Kubernetes, Mesos or any platform that supports Docker containers. Deployment Pipeline can be based on Spinnaker or Jenkins for Enterprise Continuous Delivery. AutoPilot CV stage can be integrated with Jenkins pipeline using a wrapper to AutoPilot CV API. In Spinnaker AutoPilot CV integrated natively using ACA stage of Spinnaker. It can also be integrated in pipeline using Webhook stage in Spinnaker. The whole process is illustrated in the image below: