Skip to content

OpenTelemetry

OpenTelemetry (OTel) is the vendor-neutral observability framework that DeepCube is built on. DeepCube is OpenTelemetry-exclusive - no proprietary agents, no vendor lock-in. Your instrumentation works with any OTel-compatible backend.

Why DeepCube Chose OpenTelemetry

Reason What It Means for You
No proprietary agent Nothing to install beyond standard OTel SDKs
No lock-in Switch backends or run DeepCube alongside other tools
Industry standard CNCF project backed by every major cloud provider
One framework Traces, metrics, and logs through a single SDK

DeepCube + OpenTelemetry Architecture

graph TD
    subgraph "Your Application"
        A[Application Code]
        B[OTel SDK]
        C[Auto-Instrumentation]
    end

    subgraph "Optional"
        D[OTel Collector]
    end

    A --> B
    C --> B
    B -->|Direct| E[DeepCube - otlp.deepcube.ai]
    B -->|Via Collector| D
    D --> E
    E --> F[3D / Web / Studio]

You can export directly to DeepCube or route through an OpenTelemetry Collector for batching, filtering, or multi-backend fan-out.

DeepCube Configuration

All you need is the OTLP endpoint and your API key:

Setting Value
Endpoint https://otlp.deepcube.ai
Protocol OTLP/gRPC (port 443) or OTLP/HTTP
Auth Header api-key: YOUR-API-KEY 
# Environment variables - works with any OTel SDK
export OTEL_EXPORTER_OTLP_ENDPOINT=https://otlp.deepcube.ai
export OTEL_EXPORTER_OTLP_HEADERS="api-key=YOUR-API-KEY"

For per-language setup, see the Instrumentation Guides.

Language Support

Language Traces Metrics Logs Auto-Instrumentation
.NET Stable Stable Stable Yes
Java Stable Stable Stable Yes
Python Stable Stable Stable Yes
Node.js Stable Stable Experimental Yes
Go Stable Stable Stable Limited

Key Concepts

OTLP Protocol

OpenTelemetry Protocol (OTLP) is the wire format DeepCube natively supports:

  • gRPC (port 4317) - Best performance, binary encoding
  • HTTP/protobuf (port 4318) - Firewall-friendly alternative

DeepCube accepts both on port 443 with TLS.

Sampling

Not every request needs to be traced. OTel supports:

  • Head sampling - Decide at request start (ParentBased, TraceIdRatio)
  • Tail sampling - Decide after the trace completes (via Collector)

For DeepCube throughput limits by plan, see Plans & Pricing.

Further Reading

Next Steps