Skip to content

Full-Stack Observability

In a distributed cloud environment, "Is it running?" is the wrong question. The enterprise question is "How well is it running?" and "If latency spikes, where is the bottleneck?"

Cyberun Cloud avoids AGPL-licensed components. We built a "Glass-Box" observability stack powered by OpenSearch, ensuring your monitoring data enjoys complete sovereignty and compliance.

Unified Telemetry Architecture

1. Metrics: Infrastructure Heartbeat

  • Source: Integration with the OpenSearch ecosystem.
  • Aggregation: Metrics are shipped via Remote Write to OpenSearch for long-term retention.
  • Visualization: Out-of-the-box OpenSearch Dashboards provide "Golden Signal" (Latency, Traffic, Errors, Saturation) visibility.

2. Logs: Centralized Auditing

  • Ingestion: Our pipeline (Fluent Bit) automatically scrapes container stdout/stderr.
  • Storage: Data is aggregated into the OpenSearch cluster. OpenSearch offers superior full-text search and aggregation analytics.
  • Capabilities: Supports complex structured queries (PPL/DSL) and pattern-based anomaly detection.

3. Traces: End-to-End Analysis

  • Integration: Supports the OpenTelemetry (OTEL) standard.
  • Value: Using OpenSearch Trace Analytics, you can trace a single HTTP request from the Tokyo Load Balancer to the New York database, pinpointing latency hotspots with millisecond precision.

Multi-Tenant Isolation

In Cyberun’s multi-tenant environment, your observability data remains strictly private.

  • Index Isolation: Each tenant has dedicated OpenSearch Index Patterns.
  • RBAC: Leveraging OpenSearch's fine-grained security plugin, we ensure you can only query logs belonging to your namespaces.

Intelligent Alerting (AIOps)

  • Pre-built Rules: We ship with hundreds of alerts based on SRE best practices (e.g., KubeNodeNotReady, HighErrorRate).
  • Multi-Channel Notification: Alerts are routed via Alertmanager to Slack, Email, or Webhooks.