How KubeVela Enables Full‑Stack Declarative Observability for Cloud‑Native Apps

Introduction

KubeVela is an out‑of‑the‑box modern application delivery and management platform. This article focuses on KubeVela’s observability system, introducing the challenges of cloud‑native observability and KubeVela’s solutions.

Cloud‑Native Observability Challenges

As cloud‑native workloads migrate to Kubernetes, developers face two major challenges: difficulty standardizing diverse workloads, leading to fragmented platform layers, and high complexity that raises usage barriers and stability risks.

Traditional monitoring relies on rigid scripts that bind infrastructure to observability tools, creating monolithic pipelines for metrics collection, log aggregation, and alerting.

Figure 1: CNCF landscape plugins exceed 1000+

Additional challenges include data silos across business units, multi‑cloud/multi‑cluster configurations, and disconnected data pipelines between exporters and dashboards, leading to wasted storage and delayed or inaccurate metrics.

Figure 2: Mismatch between dashboards and exporters

Full‑Stack Declarative Observability

Inspired by the declarative API philosophy of cloud‑native, KubeVela proposes “Full Stack Observability as Code”, unifying configuration of monitoring probes, collection, storage, visualization, and alerting across environments.

Open‑source projects such as Prometheus, Grafana, and OpenTelemetry have become de‑facto standards, making unified observability feasible.

KubeVela’s Prism sub‑project implements the Kubernetes Aggregated API Server (AA) pattern to translate third‑party APIs (e.g., Grafana) into native Kubernetes resources. Users can query Grafana dashboards via kubectl get grafanadashboard, with Prism handling the conversion.

Figure 3: Prism bridges native Kubernetes API and Grafana

Prism supports three deployment modes (self‑hosted, KubeVela plugin, cloud service) and uses Grafana’s own storage as the source of truth, ensuring consistent state across Kubernetes and Grafana interfaces. Access control leverages Kubernetes RBAC, eliminating data leakage concerns.

Orchestrating Observability with Application Lifecycle

KubeVela’s lightweight workflow engine enables DAG‑based orchestration of observability resources, tying them to application creation, updates, and deletion without extra pod overhead.

Use CUE to define custom Definition files that describe metrics collection, log aggregation, data source creation, and dashboard import.

Platform engineers provide reusable Traits or WorkflowSteps that developers bind to their applications.

Key questions addressed:

How to flexibly compose declarative observability APIs into the application lifecycle?

How to extend and customize observability for diverse scenarios on demand?

Design Principles

Workflow‑centric delivery model with a DAG engine that manages resources and supports multi‑cluster, multi‑cloud environments.

Programmable, infrastructure‑as‑code approach using CUE for complete declarative specifications.

Multi‑Cluster / Hybrid‑Cloud Unified Observability

KubeVela natively supports multi‑cluster and hybrid‑cloud deployments, allowing per‑cluster observability configurations.

Figure 6: Differentiated observability across clusters with unified delivery

Out‑of‑the‑Box Observability Experience

Users can enable observability addons with a single vela addon enable command, supporting multi‑cluster installation and selective deployment.

Figure 7: Installing observability addons across clusters

KubeVela provides built‑in dashboards for:

Kubernetes system metrics (cluster health, API server load, etcd usage).

KubeVela controller performance.

Application‑level views for Deployments and Nginx logs.

Figure 8: Kubernetes system dashboard

Figure 9: KubeVela controller dashboard

Figure 10: Application visualization

Figure 11: Deployment monitoring dashboard

Figure 12: Nginx log analysis dashboard

Customizing Observability Pipelines

Beyond built‑in addons, users can extend observability by defining CUE‑based workflow steps, creating RecordingRules or AlertingRules for Prometheus, and leveraging KubeVela Pipeline to orchestrate complex setups.

Figure 13: Automatic Grafana dashboard creation after KubeVela application deployment

Figure 14: Using KubeVela Pipeline to orchestrate observability infrastructure

Future Directions

KubeVela will expand its observability suite to include application probes, further integrate Dashboard‑as‑Code concepts, and continue to simplify the bridge between application models and monitoring, reinforcing stability and developer productivity in cloud‑native environments.