Network Tracing

Measure API requests and correlate with backend traces.

What to Capture (OTel-Compatible Names)

Attribute	OTel Name	Purpose
Method	`http.request.method`	GET, POST, etc.
Status	`http.response.status_code`	Success/failure
URL	`url.path`	Endpoint (sanitized)
Duration	`http.request.duration`	Request time (ms)
Size	`http.response.body.size`	Payload bytes

Using OTel naming = easier migration when OTel mobile matures. See references/otel-mobile.md for rationale.

Distributed Tracing

Propagate trace context to backend:

Client Request
    │
    ├── traceparent: 00-{trace_id}-{span_id}-01
    ├── X-Request-Id: {uuid}
    └── X-Session-Id: {session}
         │
         ▼
    Backend (correlates logs with trace_id)

Key Thresholds

Metric	Good	Acceptable	Poor
API p50	<500ms	<1s	>1s
API p95	<2s	<5s	>5s
Error rate	<1%	<3%	>3%

Integration Options

Choose based on existing vendor:

Vendor	iOS	Android	Approach
Sentry	Auto URLSession swizzling	OkHttp integration	Automatic
Datadog	URLSession delegate	OkHttp interceptor	Semi-auto
Embrace	Auto-instrumentation	Auto-instrumentation	Automatic
Custom	Manual interceptor	Manual interceptor	Manual

Automatic (swizzling): Less code, may miss custom clients Manual (interceptors): More control, works with any HTTP client

Platform Integration Points

Platform	Manual Option	Works With
iOS	URLSession delegate	All URLSession-based clients
iOS	Alamofire EventMonitor	Alamofire
Android	OkHttp Interceptor	OkHttp, Retrofit
Android	Ktor HttpClientPlugin	Ktor
RN	fetch wrapper	Native fetch
RN	axios interceptor	axios

Implementation

See references/mobile-challenges.md (Client-Backend Correlation) for:

W3C trace header format
Platform-specific interceptor code
Backend log correlation patterns

See references/performance.md (Network section) for latency budgets.

Search AI Tools

network-tracing

Install this agent skill to your Project

SKILL.md