avuru obs vs. Prometheus
Prometheus is the de-facto standard for metrics — pull-based scraping, PromQL, alerting and an enormous exporter ecosystem. It is excellent at what it does. avuru obs takes a different angle: RED metrics derived automatically and stored next to your traces and logs for correlation, rather than as a separate metrics silo.
:::note Honest status Metrics (RED) are on the avuru obs roadmap (M3). If metrics are your primary need today, Prometheus remains the stronger choice — and the two run well together (see below). :::
At a glance
| Prometheus | avuru obs | |
|---|---|---|
| Scope | Metrics | Traces, logs, metrics, profiling |
| Collection | Pull / scrape (+ exporters) | eBPF RED + OTLP push |
| Query | PromQL | Correlated queries across signals |
| Long-term storage | Add Thanos / Mimir / Cortex | Built into ClickHouse |
| High cardinality | Can get expensive | Columnar store designed for it |
| Correlation with traces/logs | External | Native — shared trace_id |
| License | Apache-2.0 | AGPL-3.0 |
Why teams look at avuru obs
- RED without scrape configs. Rate, errors and duration come from eBPF automatically, so you get golden-signal metrics per service with no exporter wiring. See Architecture.
- Metrics that link to the trace. A latency spike pivots straight to the slow traces and their logs, because everything is one store.
- One engine for retention. No separate long-term-storage layer to bolt on.
How they fit together
You don't have to choose immediately. Prometheus can remote-write into an OTLP pipeline, and avuru obs ingests OTLP — so you can feed existing Prometheus metrics into avuru while you evaluate, and keep PromQL where you need it. See the OTLP bridge guide. In the other direction, avuru's alert webhooks can feed your existing Alertmanager routing tree, so avuru-detected outages ride your dedup and silences.
When Prometheus is the better fit
- You live in PromQL and Alertmanager and rely on that ecosystem.
- You need the breadth of the exporter catalog for infra/third-party metrics.
- You need mature, independently scalable metrics today rather than on the roadmap.
:::tip Try it side by side Install in 30 seconds — the eBPF RED metrics show up with zero scrape configuration. :::