avuru obs vs. DeepFlow
DeepFlow is a mature, deep eBPF observability platform from Yunshan Networks. It shares a lot of DNA with avuru obs: both collect telemetry zero-code with eBPF and both store every signal in ClickHouse. The differences are about packaging and scope — avuru obs ships a complete first-party UI and adds error tracking, while DeepFlow brings deeper in-agent protocol and network-flow analysis and a permissive Apache-2.0 core.
At a glance
| DeepFlow | avuru obs | |
|---|---|---|
| Collection | Zero-code eBPF (+ cBPF) agent | Zero-code eBPF via upstream OBI |
| Storage | ClickHouse (+ SmartEncoding tags) | Single ClickHouse engine for all signals |
| UI | Grafana + DeepFlow Grafana apps (CE) | Complete first-party UI (map, traces, logs, RED) |
| Protocol parsing | Broad in-agent L7 decoders + Wasm plugins | Leans on OBI for protocol parsing |
| Error tracking | Not a dedicated feature | Native error tracking + Sentry-SDK ingest |
| Ingest | OTLP, plus its own agent pipeline | Native OTLP (:4318/:4317) |
| Deploy | Agent + server + ClickHouse + MySQL + Grafana | One Helm chart |
| License | Apache-2.0 (core) | AGPL-3.0 |
Why teams choose avuru obs
- A UI in the box. DeepFlow's community edition renders its data through Grafana and DeepFlow's Grafana apps; you assemble and maintain that layer. avuru obs ships its own service map, trace viewer, flamegraph, trace compare, logs and RED dashboards — no Grafana to wire up. See Architecture.
- Errors are a first-class signal. avuru obs adds native error tracking with Sentry-SDK-compatible ingest, so the SDKs your apps already use can send exceptions straight into the same store as your traces and logs. DeepFlow focuses on eBPF metrics, tracing, flow logs and profiling and has no dedicated error-tracking product.
- Fewer moving parts. DeepFlow runs an agent, a server, ClickHouse, a MySQL metadata database and Grafana. avuru obs is one Helm chart over a single ClickHouse store.
- Connection health on the map. Per-edge TCP RTT and failed connections from OBI's kernel stats surface directly on the service map — see Network health.
How they coexist
DeepFlow can export over OTLP, and avuru obs is an OTLP-native backend, so you don't have to choose blind:
- Install avuru obs alongside DeepFlow.
- Point DeepFlow's OTLP export (or your shared OTel Collector) at the avuru
gatewayon:4318/:4317, keeping DeepFlow's own pipeline running. - Compare the correlated view and the first-party UI against DeepFlow's Grafana dashboards, then decide what each tool is best at.
See the OTLP bridge guide for exporter settings.
When DeepFlow is the better fit
- You want DeepFlow's deep network-flow observability and its broad set of in-agent L7 protocol decoders — avuru obs leans on OBI and covers fewer protocols today.
- You prefer a permissive Apache-2.0 license over avuru obs's copyleft AGPL-3.0.
- You're happy building views in Grafana and want its dashboarding flexibility, or you rely on DeepFlow's AutoTagging and SmartEncoding at scale.
- You need a broad, battle-tested platform today — avuru obs is young, and some capabilities are still landing. Check Feature status before committing.
:::tip Try it side by side Install in 30 seconds and export one service over OTLP — you'll see the first-party map, traces and logs correlated in one UI. :::