Skip to main content

Overview

KubePattern is an innovative static analysis tool for Kubernetes manifests that goes beyond traditional linting, focusing on recognizing and suggesting cloud-native architectural patterns.

warning

KubePattern is also my Bachelor Degree Thesis Project and is currently under development. It will be fully open source on march 2026.

Have a look and stay tuned if you have further question please contact me.

A special thanks to Sigemi and for the knoledge the tools and the opportunity to build KubePattern during my internship.

What is KubePattern?​

KubePattern is a pattern-oriented validation tool that analyzes Kubernetes configurations to identify opportunities for applying architectural patterns and detect configuration "smells" that deviate from community best practices and corporate policies.

Key Features​

Pattern-as-Code​

KubePattern adopts a declarative approach where patterns are defined through external JSON files, enabling:

  • Extensibility: Add new patterns without modifying the code
  • Customization: Define patterns specific to corporate policies
  • Versioning: Manage pattern evolution over time

Architectural Pattern Recognition​

Unlike traditional linters that only verify field presence, KubePattern:

  • Identifies complex patterns such as Sidecar, Health Probe, Predictable Demands
  • Analyzes relationships between multiple resources
  • Provides contextual suggestions with reference documentation

Scoring and Confidence System​

Each detected pattern includes:

  • Confidence Level: Probability that the pattern has been correctly identified (LOW/MEDIUM/HIGH)
  • Severity: Criticality level (INFO/WARNING/CRITICAL)
  • Category: Application scope (Reliability/Security/Performance)
  • Scores: Detailed scores for each performed check

Architecture​

KubePattern is developed in Java and uses:

  • Official Kubernetes Java Client: Library used to read and apply resources on the cluster
  • JsonPath: Library used to interpret pattern definitions (JSON)
  • Custom Resource Definitions (CRD): Results are exposed as native Kubernetes resources

The tool creates Custom Resources (K8sPattern) that represent identified patterns, making results easily integrable with other Kubernetes tools.

Output and Reporting​

Analysis results are available as Kubernetes CRDs

Custom Resource Definition (CRD)​

apiVersion: sigemi.it/v1
kind: K8sPattern
metadata:
  name: sidecar-pattern-0i9j-8k7l6m5n4o3p
  namespace: pattern-analysis-ns
spec:
  name: Sidecar
  type: Structural
  severity: INFO
  category: Reliability
  confidence: HIGH
  scores:
    shared-volume-mount: 60
    similar-replica-count: 40
  resources:
    - name: main-app
      namespace: development
      uid: 9a7b6c5d-4e3f-2g1h-0i9j-8k7l6m5n4o3p
    - name: fluent
      namespace: development
      uid: 1b2c3d4e-5f6g-7h8i-9j0k-1l2m3n4o5p6q

Next Steps​