Managing and Monitoring Kubernetes
Kubernetes is an open-source platform that focuses on automating container management in a flexible and dynamic way. Its various parts all build on one another to orchestrate an entire microservices environment. Kubernetes groups containers running on the same host machine into pods and assigns key-value pair tags for identifying and sorting them. Pods then execute on service requests from either human users or other environment components acting as clients. Internal controllers constantly match the actual system state with the desired state called for in the configuration, and subsequently clone, distribute, and run pods to produce that desired state. For more details, please see the Kubernetes documentation.
After deployment into a client environment, Instana will automatically map out the entire infrastructure and identify how the different technologies are interacting with one another with its Dynamic Graph back-end model.
Kubernetes Performance and Configuration Metrics
The Instana agent Kubernetes sensor is preconfigured to understand common baselines for proper Kubernetes behavior within a given environment. Instana will automatically begin collecting metrics on Kubernetes’ performance in container management and configured state maintenance.
The effectiveness of resource management is captured in typical metrics such as:
- CPU Usage
- Memory Usage
- Block IO
Instana also collects specific Kubernetes configurations for display in the UI dashboards. The agent sensor will automatically identify different Kubernetes components such as pods and containers, indexing them and making them searchable by users directly from the same dashboards that display the resource management performance.
Please see Instana’s documentation for more details on installing Instana into a Kubernetes platform.
Kubernetes Health Monitoring and Management
In addition to monitoring Kubernetes’ performance metrics, the Instana agent sensor will also collect health metrics on the Kubernetes platform itself. The health state is displayed in the UI in real-time as healthy, sick, or failing. In a containerized environment there are a lot of moving parts that can break down. The Kubernetes API server’s HTTP connection could break down, pods can get mistagged causing resources to get mismanaged, or user error could cause a scheduler to get unplugged. Instana’s real-time health monitoring allows DevOps engineers to ensure their valuable Kubernetes platforms manage their microservices environments smoothly.