Skip to main content

​
Overview

All trace data is sent asynchronously in the background. Your application code is never blocked waiting for Artanis. 
# This returns immediately (~0.1ms overhead)
trace.output(response)

# Data is sent automatically in the background

​
How It Works

When you call trace.input(), trace.output(), or trace.state():
  1. Data is serialized immediately (< 0.1ms)
  2. Data is sent asynchronously in the background
  3. Your code continues without waiting
Network calls happen in the background and never block your application.

​
Performance Impact

The SDK adds minimal latency to your application:
OperationLatency
trace.input(), trace.output(), etc.< 0.1ms (serialization only)
Network calls0ms (async, non-blocking)
P50 overhead< 0.05ms per trace operation
P99 overhead< 0.5ms per trace operation
No synchronous network calls means zero waiting for responses.

​
Reliability

If network fails, traces are dropped silently:
  • No exceptions thrown - your application never crashes due to tracing
  • No retries - prevents memory leaks in long-running applications
  • No blocking - network issues never slow down your code
See the Error Handling guide for more details.

​
Next Steps

Initialization

Configure the SDK

Error Handling

Understand how the SDK handles errors