OpenTelemetry-Native APM Tools That Aren't Datadog

Five APM tools that use OpenTelemetry as the primary path, not as a bolt-on. Compared on what 'native' actually means in practice.

Lhoussine
May 9, 2026·7 min read

OpenTelemetry-Native APM Tools That Aren't Datadog

If you want to avoid vendor lock-in, the data layer matters more than the destination. Five tools that use OpenTelemetry as the primary path — not as a bolt-on adapter.

For deeper context see Datadog vs OpenTelemetry: when to drop the agent.

What "native" actually means

Three properties distinguish native tools from accepts-OTLP tools:

  1. The default SDK is OpenTelemetry. Your npm install is @opentelemetry/sdk-node or a wrapper around it, not a vendor-specific tracer.
  2. Dashboards and queries assume OTel attribute names. http.target, client.address, enduser.id — the standard semantic conventions, not vendor-renamed equivalents.
  3. Migration to a different OTel backend is a config flag. Change one URL, your data goes elsewhere.

By that strict definition, the following tools qualify.

1. SecureNow

The npm package wraps @opentelemetry/sdk-node. Auto-instruments Express, Next.js, NestJS, Fastify, Koa, Hono, Hapi, Restify, Polka, Feathers, h3. Adds the firewall and AI investigation as additional layers, all using OTel data.

Pricing: $5/TB scanned, free tier 1 GB/month.

Best for: Node-heavy SaaS that wants APM + security in one tool.

Caveat: Currently Node-only. Python and Go on roadmap.

See the comparison page and APM + security in one tool.

2. SigNoz

ClickHouse-backed observability with OpenTelemetry as the primary SDK. Self-host or cloud. Mature feature set.

Pricing: Self-host free + infrastructure cost (~$300/month for moderate workload). Cloud at $0.30/GB ingested.

Best for: Teams that want self-hostable, OpenTelemetry-first observability with APM + logs + traces unified.

Caveat: Self-hosting requires ClickHouse expertise. Cloud version is competitive but not the cheapest.

3. Grafana Cloud (Tempo + Loki + Mimir)

The Grafana stack: Tempo for traces, Loki for logs, Mimir for metrics. All OpenTelemetry-compatible.

Pricing: Free 50 GB tier, $0.50/GB beyond.

Best for: Teams already on Grafana, large multi-language stacks.

Caveat: Three products, three pricing axes. Can be confusing at scale.

4. Dash0

Predictable per-GB pricing on a clean OTel-native UI. Newer entrant focused on cost predictability.

Pricing: $0.30/GB ingested with 5 GB free.

Best for: Teams burned by Datadog's variable bills who want a predictable per-byte cost.

Caveat: Smaller community, fewer integrations than mature alternatives.

5. HyperDX

OpenTelemetry-native with strong frontend session-replay correlation. Built on ClickHouse.

Pricing: Open-source self-host free, cloud $0.40/GB.

Best for: Teams where frontend ↔ backend correlation matters, like e-commerce or content platforms.

Caveat: Smaller ecosystem of integrations than Grafana or SigNoz.

What you give up vs Datadog

Three classes of features that OpenTelemetry-native tools don't yet match:

1. Continuous profiler (Datadog Pyroscope-equivalent). Some OTel tools have this; not all. Check before assuming.

2. Network Performance Monitoring (NPM). Datadog's eBPF-based network monitoring. Not natively in OTel; needs a separate tool (Cilium Hubble, Pixie).

3. Cloud Workload Security (CWP) / CSPM. Filesystem and process-level security monitoring at the host. Out of scope for OTel-native tools; needs a separate product like Wiz, Aqua, or Sysdig.

For most teams, none of these matter. For some specific use cases, they're deal-breakers.

The migration logic

If you're starting from Datadog and want to be OTel-native:

Step 1: Replace dd-trace with @opentelemetry/sdk-node or SecureNow. Keep Datadog as the destination via OTLP.

Step 2: Verify dashboards still work. Datadog's UI supports OTLP-shaped traces; some attributes will rename (http.target instead of http.url, etc.) but core functionality is preserved.

Step 3: Stand up a parallel destination — SigNoz, SecureNow, Grafana Tempo. Send the same data to both.

Step 4: Migrate dashboards and alerts. This is the long pole.

Step 5: Cancel Datadog when ready.

The point is: with OTel as the data layer, you can do step 5 whenever you want. Without it, step 5 requires re-instrumenting the entire fleet.

The recommendation matrix

SituationPick
Self-host, want full controlSigNoz
Node-heavy SaaS, want consolidated APM + securitySecureNow
Already on GrafanaGrafana Cloud
Want predictable per-GB pricingDash0
Frontend correlation mattersHyperDX
Need continuous profiler(stay on Datadog or add Pyroscope)

Whichever you pick, the data layer is OpenTelemetry — and that's the only decision that's reversible.

Related

Frequently Asked Questions

What does 'OpenTelemetry-native' mean?

Tools where OpenTelemetry is the primary instrumentation path — not just an accepted format. The SDK, dashboards, alerts, and detection rules all assume OTel data shapes.

Doesn't Datadog accept OTLP?

Yes, but the path of least resistance is still the dd-trace agent. Their dashboards and queries assume Datadog's internal trace format. OTel is supported but not native.

Why does it matter?

Lock-in. With OTel-native, switching backends is a config change. With proprietary-first, switching is a re-instrumentation.

What's the catch with OTel-native tools?

Some Datadog-only features (continuous profiler, NPM, certain integrations) don't have OTel equivalents. If you depend on those, you're stuck with the proprietary tier.

Recommended reading

What 1.2B Requests Look Like: Anomaly Patterns from the SecureNow Firewall Fleet

Aggregated, anonymized data from 1.2B requests across the SecureNow customer fleet. Top anomaly types, peak hours, and the day-of-week patterns nobody publishes.

May 9
10 Best Application Security Monitoring Tools in 2026

An honest, side-by-side comparison of the ten most-deployed application security monitoring tools — from enterprise platforms to free open-source options.

May 9
The 2026 npm Supply-Chain Attack Survey, Q2

A quarterly tally of malicious npm packages, the major incidents, and detection patterns. April 2026 set a new record at 847 confirmed malicious packages — here's what they did and how to detect them.

May 9