NetMediate Diagnostics

Built-in OpenTelemetry traces and metrics

NetMediate emits diagnostics primitives for every mediator operation via the standard .NET ActivitySource and Meter APIs.

ActivitySource

Name: NetMediate (NetMediateDiagnostics.ActivitySourceName)

An Activity is started for each operation:

Operation	Activity name	Tags
IMediator.Send	NetMediate.Send	netmediate.operation, netmediate.message_type
IMediator.Request	NetMediate.Request	netmediate.operation, netmediate.message_type
IMediator.Notify	NetMediate.Notify	netmediate.operation, netmediate.message_type
IMediator.RequestStream	NetMediate.RequestStream	netmediate.operation, netmediate.message_type

Activities are only started when the ActivitySource has at least one listener — there is no overhead when tracing is not configured.

Meter

Name: NetMediate (NetMediateDiagnostics.MeterName)

Counter name	Constant	Incremented by
netmediate.send.count	NetMediateDiagnostics.SendCountMetricName	IMediator.Send
netmediate.request.count	NetMediateDiagnostics.RequestCountMetricName	IMediator.Request
netmediate.notify.count	NetMediateDiagnostics.NotifyCountMetricName	IMediator.Notify
netmediate.dispatch.count	NetMediateDiagnostics.DispatchCountMetricName	internal handler dispatch
netmediate.stream.count	NetMediateDiagnostics.StreamCountMetricName	IMediator.RequestStream

All counters carry a message_type tag with the message CLR type name. Counters are only recorded when the Counter is enabled — there is no overhead when metrics are not configured.

Quick integration example

builder.Services.AddOpenTelemetry()
    .WithTracing(t => t.AddSource(NetMediateDiagnostics.ActivitySourceName))
    .WithMetrics(m => m.AddMeter(NetMediateDiagnostics.MeterName));

For DataDog, use the dedicated package instead:

builder.Services.AddNetMediateDataDogOpenTelemetry(options =>
{
    options.ServiceName    = "my-service";
    options.OtlpEndpoint   = new Uri("http://localhost:4318");
});

See DataDog Integration for full details.

Performance comparison (main vs current branch)

Measurements were captured with the same load scenarios used in tests, using 5 runs per branch and reporting the median throughput:

• command: 20,000 operations
• request_parallel: 10,000 operations

Environment used for these measurements:

• CPU/vCPU: AMD EPYC 7763 with 4 vCPU
• Memory: 15 GiB RAM

Scenario	origin/main median throughput (ops/s)	Current branch median throughput (ops/s)	Delta
command	172,650.29	179,843.50	+4.17%
request_parallel	280,944.65	118,619.88	-57.78%

Notes:

• command throughput improved slightly with the current branch in this environment.
• request throughput is lower because this branch adds extra per-request work in the hot path (behavior-chain resolution + diagnostics counters/tags + activity lifecycle checks).
• benchmark results are sensitive to runtime environment noise. Re-run multiple times for stable medians before release decisions.

See Also

• Benchmarks — full benchmark matrix and reproduction steps
• DataDog Integration — DataDog integration guide