Closing the Kubernetes Observability Gap: Introducing Mermin

We are thrilled to introduce Mermin, an open source solution designed to provide deep, comprehensive observability of all container traffic. 

The Observability Problem: Siloed Worlds

Microservice observability is typically split into two disconnected domains: Application Performance Monitoring (APM) and Network Performance Monitoring (NPM).

• APM: Monitors MELT (metrics, events, logs, and traces) to figure out what is going on inside a deployed application and to troubleshoot issues when they arise. Most teams have or are in the process of standardizing on OTel (OpenTelemetry) as the data schema for their observability solutions. While they can see what is going on in their application, they don’t see the network status. This can often lead to the false assumption that any unexplained error or undesired behavior must be the network's fault.

• NPM: Manages the underlying network infrastructure through a mix of telemetry data, traps and flow logs. As a result of cloud services, containerization and overlay networks, network teams have a hard time identifying which application causes which traffic. Making it impossible for teams to prioritize critical applications correctly or to effectively plan for future hardware needs. They potentially see an application error, but have a hard time notifying the right team as ownership of traffic is often unclear.

The result is the infamous "blame game." Teams are correct based on their siloed data, but they lack the ability to connect a specific, slow application request to the actual network conversation that serviced it. This gap is where hours of engineering time are lost during root cause analysis.

Introducing Mermin: Ground-Truth Network Visibility

Mermin is designed to eliminate this blind spot. It is a lightweight, eBPF-based agent that runs inside your Kubernetes cluster. By attaching directly to the Linux kernel, Mermin captures ground-truth data about every network conversation between your services, requiring zero application code changes or instrumentation.

Key Features

• eBPF-based Data Collection: Mermin captures network data directly from the Linux kernel, making it incredibly fast and efficient.

• Automated Kubernetes Metadata Enrichment: It automatically correlates network flows with the Kubernetes objects involved (Pod, Deployment, Service, Namespace). This transforms cryptic IP addresses into human-readable service names, making troubleshooting intuitive.

• OpenTelemetry (OTel) Native: Mermin formats its rich network data as OpenTelemetry traces using the flow semantic convention. This crucial feature ensures Mermin integrates seamlessly into the industry-standard observability ecosystem.

• Bidirectional Flow Analysis: Unlike simple network metrics, Mermin captures data for both directions of a conversation (client-to-server and server-to-client), essential for fully understanding the nature of the traffic.
  

How Mermin Transforms Your XOps Workflow

Mermin empowers teams across the organization, providing true full-stack visibility by extending the MELT framework (Metrics, Events, Logs, Traces) to include Flows.

Get Started with the Mermin

Mermin supports common Kubernetes distributions (EKS, GKE, AKS, OpenShift), providing a consistent visibility solution across diverse environments.

We invite you to experience the ground-truth network visibility Mermin offers and help us refine it. This is your chance to provide feedback that will directly shape the future of container observability!

Click here to sign up for access to the Beta.