avuru obs vs. the Grafana stack
The Grafana LGTM stack — Loki (logs), Grafana (dashboards), Tempo (traces), Mimir (metrics), plus Pyroscope for profiling — is powerful and flexible, with a huge dashboarding ecosystem. The trade-off is that you assemble and operate several systems, each with its own store to size, scale and back up. avuru obs collapses that into one engine and one UI.
At a glance
| Grafana stack (LGTM) | avuru obs | |
|---|---|---|
| Components | Loki + Tempo + Mimir + Grafana (+ Pyroscope) | One hub + gateway + ClickHouse + UI |
| Stores to operate | One per signal | One (ClickHouse, all signals) |
| Instrumentation | Agents/SDKs per signal (Alloy, OTel) | eBPF auto-discovery + OTLP |
| Correlation | Configured across data sources | Built in — same store, shared keys |
| Dashboards | Best-in-class, huge ecosystem | Purpose-built UI (less flexible today) |
| Ingest | OTLP, Prometheus, Loki push, etc. | Native OTLP (:4318/:4317) |
| License | AGPLv3 (Grafana) / Apache-2.0 (components) | AGPL-3.0 |
Why teams switch
- One system instead of four. Sizing, upgrading and backing up a single ClickHouse cluster is far less work than operating Loki, Tempo and Mimir separately. See Architecture.
- Correlation without wiring. Traces, logs (and soon metrics/profiles) share a store, so a trace → logs pivot is a query — not a cross-data-source link you configure and maintain.
- The map is free. eBPF discovery gives you a live service map with no agent config per signal.
How to migrate
The Grafana stack is OTLP-friendly, so you can move incrementally:
- Install avuru obs alongside your stack.
- Add the avuru
gatewayas an OTLP exporter in Grafana Alloy / your OTel Collector, keeping your existing exporters so both receive data. - Validate traces and logs in avuru obs, then peel off Loki/Tempo (and Mimir once metrics land) as you're ready.
See the OTLP bridge guide for collector settings.
When the Grafana stack is the better fit
- You rely on Grafana's dashboarding flexibility and plugin ecosystem.
- You need mature, independently scalable metrics (Mimir) today — avuru's metrics are RED-focused and still landing (M3).
- You've already invested in operating the stack and it meets your needs.
:::tip Try it side by side Install in 30 seconds, dual-export from Alloy, and compare the correlated view against your current data sources. :::