Features
Observability that adapts to how you work
From a single binary on your laptop to a federated mesh across continents — Observ scales with your organization without changing the tool.
Standalone mode
Full observability on your laptop. Zero infrastructure.
Observ runs as a single binary — no Docker, no Kubernetes, no external dependencies. Download it, point your app at it via OTLP, and get logs, metrics, and traces in a local dashboard. Perfect for development, debugging, and testing your instrumentation before pushing to production.
- Single binary, under 50 MB
- Starts in under 2 seconds
- Local Vortex + SQLite storage
- Full OTLP ingestion (gRPC & HTTP)
- Same query engine as production
Root Cause Analysis
From alert to diagnosis in seconds, not hours.
When an incident fires, Observ automatically correlates logs, metrics, and traces across services to build a timeline. Then an LLM analyzes the context and produces a human-readable diagnosis — explaining what happened, why, and what to do about it. No more jumping between 15 tabs to find the root cause.
- Automatic cross-signal correlation
- LLM-generated incident summary
- Root cause hypothesis with confidence score
- Suggested remediation steps
- Accessible to non-experts (support, management)
Federation
Multiple instances, one query. Data stays where it lives.
Large organizations run multiple Observ instances — one per team, per region, or per business unit. Federation connects them via Arrow Flight protocol in a star topology: a central "Chef" instance can query any "Member" instance in real-time, with full SQL push-down. Each member keeps its own data, syncs metadata to the Chef, and operates independently if the network goes down.
- Star topology: Chef + Members
- Arrow Flight wire protocol (port 4316)
- Full SQL push-down to members
- Metadata sync for cross-instance discovery
- Each member operates independently if disconnected
- Bearer token authentication with expiry
Edge Collector
Local detection, without flooding the network.
Deploy a headless Observ instance (no UI) directly inside your Kubernetes clusters. It ingests OTLP data locally, stores it in compact storage, and runs root cause analysis on the spot — without sending every metric and trace to the central platform. Only alerts and diagnosis results are forwarded. The central platform can still query edge data on-demand via federation when deep investigation is needed.
- Headless mode (no UI, minimal footprint)
- Local OTLP ingestion and storage
- On-device root cause analysis
- Only alerts and summaries sent upstream
- Queryable on-demand via federation
- Dramatic reduction in network egress costs
See it in action
Download the free standalone version and start exploring in minutes. The intelligence engine ships as the Intelligence add-on; multi-cluster federation is part of Enterprise.