Multiple mechanisms determine whether spans generated by your applications are sent to Datadog (ingested). The logic behind these mechanisms is in the SDKs and in the Datadog Agent. Depending on the configuration, all or some of the traffic generated by instrumented services is ingested.

Each ingested span has a unique ingestion reason that refers to one of the mechanisms described on this page. The usage metrics datadog.estimated_usage.apm.ingested_bytes and datadog.estimated_usage.apm.ingested_spans are tagged by ingestion_reason.

Use the Ingestion Reasons dashboard to investigate each ingestion reason in context and identify which configuration options to focus on.

Head-based sampling

The default sampling mechanism is called head-based sampling. The decision to keep or drop a trace is made at the start of the root span and then propagated to other services as part of their request context (for example, as an HTTP request header).

Because the decision is made at the beginning of the trace and conveyed to all parts, the trace is kept or dropped as a whole.

You can set sampling rates for head-based sampling in two places:

• At the Agent level (default)
• At the SDK level: any SDK mechanism overrides the Agent setup.

In the Agent

ingestion_reason: auto

The Datadog Agent continuously sends sampling rates to SDKs to apply at the root of traces. The Agent adjusts rates to achieve a target of ten traces per second overall, distributed to services depending on the traffic.

For instance, if service A has more traffic than service B, the Agent might vary the sampling rate for A such that A keeps no more than seven traces per second, and similarly adjust the sampling rate for B such that B keeps no more than three traces per second, for a total of 10 traces per second.

Remote configuration

Sampling rate configuration in the Agent is configurable remotely if you are using Agent version 7.42.0 or higher. To get started, set up Remote Configuration and then configure the ingestion_reason parameter from the Ingestion Control page. Remote Configuration allows you to change the parameter without restarting the Agent. Remotely set configuration takes precedence over local configurations, including environment variables and settings from datadog.yaml.

Local configuration

Set the Agent’s target traces-per-second in its main configuration file (datadog.yaml) or as an environment variable:

@param target_traces_per_second - integer - optional - default: 10
@env DD_APM_TARGET_TPS - integer - optional - default: 10

Notes:

• The traces-per-second sampling rate set in the Agent only applies to Datadog SDKs. It has no effect on other SDKs such as OpenTelemetry SDKs.
• The target is not a fixed value. In reality, it fluctuates depending on traffic spikes and other factors.

Spans from traces sampled by the Datadog Agent’s automatic sampling rates are tagged with ingestion reason auto. The ingestion_reason tag is also set on usage metrics. Services using this default mechanism are labeled Automatic in the Ingestion Control Page Configuration column.

In SDKs: user-defined rules

ingestion_reason: rule

For more granular control, use SDK sampling configuration options:

• Set a specific sampling rate to apply to the root of the trace by service or resource name, overriding the Agent’s default mechanism.
• Set a rate limit on the number of ingested traces per second. The default rate limit is 100 traces per second per service instance. When using the Agent default mechanism, the rate limiter is ignored.

Note: Sampling rules are also head-based sampling controls. If the traffic for a service is higher than the configured maximum traces per second, then traces are dropped at the root. It does not create incomplete traces.

The configuration can be set by environment variables or directly in the code:

# using system property
java -Ddd.trace.sampling.rules='[{"service": "my-service", "resource": "GET /checkout", "sample_rate":1},{"service": "my-service", "sample_rate":0.2}]' -javaagent:dd-java-agent.jar -jar my-app.jar

# using environment variables
export DD_TRACE_SAMPLING_RULES='[{"service": "my-service", "resource":"GET /checkout", "sample_rate": 1},{"service": "my-service", "sample_rate": 0.2}]'

export DD_TRACE_SAMPLING_RULES='[{"service": "my-service", "resource": "GET /checkout", "sample_rate": 1},{"service": "my-service", "sample_rate": 0.2}]'

export DD_TRACE_SAMPLE_RATE=0.1
export DD_TRACE_SAMPLING_RULES='[{"service": "my-service", "sample_rate": 0.5}]'

export DD_TRACE_SAMPLING_RULES='[{"service": "my-service", "resource": "GET /checkout", "sample_rate": 1},{"service": "my-service", "sample_rate": 0.2}]'

tracer.init({
    ingestion: {
        sampler: {
            sampleRate: 0.1,
            rules: [
                { sampleRate: 0.5, service: 'my-service' }
            ]
        }
    }
});

export DD_TRACE_SAMPLE_RATE=0.1
export DD_TRACE_SAMPLING_RULES='[{"service": "my-service", "resource":"GET /checkout", "sample_rate": 1},{"service": "my-service", "sample_rate": 0.2}]'

export DD_TRACE_SAMPLE_RATE=0.1
export DD_TRACE_SAMPLING_RULES='[{"service": "my-service", "sample_rate": 0.5}]'

#using powershell
$env:DD_TRACE_SAMPLE_RATE=0.1
$env:DD_TRACE_SAMPLING_RULES='[{"service": "my-service", "sample_rate": 0.5}]'

#using JSON file   
{
    "DD_TRACE_SAMPLE_RATE": "0.1",
    "DD_TRACE_SAMPLING_RULES": "[{\"service\": \"my-service\", \"resource\": \"GET /checkout\", \"sample_rate\": 0.5}]"
}

Note: All the spans from a trace sampled using an SDK configuration are tagged with the ingestion reason rule. Services configured with user-defined sampling rules are marked as Configured in the Ingestion Control Page Configuration column.

Error and rare traces

For traces not caught by head-based sampling, two additional Datadog Agent sampling mechanisms catch critical and diverse traces that would otherwise be dropped. These samplers keep a diverse set of local traces (spans from the same host) by catching all combinations of a predetermined set of tags:

• Error traces: Error sampling provides visibility into potential system failures.
• Rare traces: Rare trace sampling maintains visibility into low-traffic services and resources across your system.

Note: Error and rare samplers are ignored for services for which you set library sampling rules.

Error traces

ingestion_reason: error

The error sampler catches pieces of traces that contain error spans not caught by head-based sampling, at a rate of up to 10 traces per second per Agent. This helps maintain visibility into errors when the head-based sampling rate is low.

With Agent version 7.33 and forward, you can configure the error sampler in the Agent main configuration file (datadog.yaml) or with environment variables:

@param errors_per_second - integer - optional - default: 10
@env DD_APM_ERROR_TPS - integer - optional - default: 10

Notes:

1. Set the parameter to 0 to disable the error sampler.
2. The error sampler captures local error traces at the Agent level. If the trace is distributed, the complete trace may not be sent to Datadog.
3. By default, spans dropped by SDK rules or custom logic such as manual.drop are excluded under the error sampler.

Datadog Agent 7.42.0 and higher

Error sampling is remotely configurable if you’re using Agent version 7.42.0 or higher. Follow the documentation to enable remote configuration in your Agents. With remote configuration, you can enable the collection of rare spans without restarting the Datadog Agent.

Datadog Agent 6/7.41.0 and higher

To override the default behavior so that spans dropped by the SDK rules or custom logic such as manual.drop are included by the error sampler, enable the feature with: DD_APM_FEATURES=error_rare_sample_tracer_drop in the Datadog Agent (or the dedicated Trace Agent container within the Datadog Agent pod in Kubernetes).

Datadog Agent 6/7.33 to 6/7.40.x

Error sampling default behavior can’t be changed for these Agent versions. Upgrade the Datadog Agent to Datadog Agent 6/7.41.0 and higher.

Rare traces

ingestion_reason: rare

The rare sampler sends a set of rare spans to Datadog. It catches combinations of env, service, name, resource, error.type, and http.status at up to 5 traces per second per Agent. This helps maintain visibility into low-traffic resources when the head-based sampling rate is low.

Note: The rare sampler captures local traces at the Agent level. If the trace is distributed, there is no guarantee that the complete trace is sent to Datadog.

Datadog Agent 7.42.0 and higher

Rare sampling is remotely configurable if you’re using Agent version 7.42.0 or higher. Follow the documentation to enable remote configuration in your Agents. With remote configuration, you can change the parameter value without restarting the Datadog Agent.

Datadog Agent 6/7.41.0 and higher

By default, the rare sampler is not enabled.

Note: When enabled, spans dropped by SDK rules or custom logic such as manual.drop are excluded under this sampler.

To configure the rare sampler, update the apm_config.enable_rare_sampler setting in the Agent main configuration file (datadog.yaml) or with the environment variable DD_APM_ENABLE_RARE_SAMPLER:

@params apm_config.enable_rare_sampler - boolean - optional - default: false
@env DD_APM_ENABLE_RARE_SAMPLER - boolean - optional - default: false

To evaluate spans dropped by SDK rules or custom logic such as manual.drop, enable the feature with: DD_APM_FEATURES=error_rare_sample_tracer_drop in the Trace Agent.

Datadog Agent 6/7.33 to 6/7.40.x

By default, the rare sampler is enabled.

Note: When enabled, spans dropped by SDK rules or custom logic such as manual.drop are excluded under this sampler. To include these spans in this logic, upgrade to Datadog Agent 6.41.0/7.41.0 or higher.

To change the default rare sampler settings, update the apm_config.disable_rare_sampler setting in the Agent main configuration file (datadog.yaml) or with the environment variable DD_APM_DISABLE_RARE_SAMPLER:

@params apm_config.disable_rare_sampler - boolean - optional - default: false
@env DD_APM_DISABLE_RARE_SAMPLER - boolean - optional - default: false

Force keep and drop

ingestion_reason: manual

The head-based sampling mechanism can be overridden at the SDK level. For example, if you need to monitor a critical transaction, you can force the associated trace to be kept. On the other hand, for unnecessary or repetitive information like health checks, you can force the trace to be dropped.

• Set Manual Keep on a span to indicate that it and all child spans should be ingested. The resulting trace might appear incomplete in the UI if the span in question is not the root span of the trace.

• Set Manual Drop on a span to make sure that no child span is ingested. The error and rare samplers are ignored in the Agent.

import datadog.trace.api.DDTags;
import io.opentracing.Span;
import datadog.trace.api.Trace;
import io.opentracing.util.GlobalTracer;

public class MyClass {
    @Trace
    public static void myMethod() {
        // grab the active span out of the traced method
        Span span = GlobalTracer.get().activeSpan();
        // Always keep the trace
        span.setTag(DDTags.MANUAL_KEEP, true);
        // method impl follows
    }
}

import datadog.trace.api.DDTags;
import io.opentracing.Span;
import datadog.trace.api.Trace;
import io.opentracing.util.GlobalTracer;

public class MyClass {
    @Trace
    public static void myMethod() {
        // grab the active span out of the traced method
        Span span = GlobalTracer.get().activeSpan();
        // Always Drop the trace
        span.setTag(DDTags.MANUAL_DROP, true);
        // method impl follows
    }
}

from ddtrace import tracer
from ddtrace.constants import MANUAL_DROP_KEY, MANUAL_KEEP_KEY

@tracer.wrap()
def handler():
    span = tracer.current_span()
    # Always Keep the Trace
    span.set_tag(MANUAL_KEEP_KEY)
    # method impl follows

from ddtrace import tracer
from ddtrace.constants import MANUAL_DROP_KEY, MANUAL_KEEP_KEY

@tracer.wrap()
def handler():
    span = tracer.current_span()
    # Always Drop the Trace
    span.set_tag(MANUAL_DROP_KEY)
    # method impl follows

Datadog::Tracing.trace(name, options) do |span, trace|
  trace.keep! # Affects the active trace
  # Method implementation follows
end

Datadog::Tracing.trace(name, options) do |span, trace|
  trace.reject! # Affects the active trace
  # Method implementation follows
end

package main

import (
    "log"
    "net/http"
    "github.com/DataDog/dd-trace-go/v2/ddtrace/ext" 
    "github.com/DataDog/dd-trace-go/v2/ddtrace/tracer"
)

func handler(w http.ResponseWriter, r *http.Request) {
    // Create a span for a web request at the /posts URL.
    span := tracer.StartSpan("web.request", tracer.ResourceName("/posts"))
    defer span.Finish()

    // Always keep this trace:
    span.SetTag(ext.ManualKeep, true)
    //method impl follows

}

package main

import (
    "log"
    "net/http"

    "github.com/DataDog/dd-trace-go/v2/ddtrace/ext"
    "github.com/DataDog/dd-trace-go/v2/ddtrace/tracer"
)

func handler(w http.ResponseWriter, r *http.Request) {
    // Create a span for a web request at the /posts URL.
    span := tracer.StartSpan("web.request", tracer.ResourceName("/posts"))
    defer span.Finish()

    // Always drop this trace:
    span.SetTag(ext.ManualDrop, true)
    //method impl follows
}

const tracer = require('dd-trace')
const tags = require('dd-trace/ext/tags')

const span = tracer.startSpan('web.request')

// Always keep the trace
span.setTag(tags.MANUAL_KEEP)
//method impl follows

const tracer = require('dd-trace')
const tags = require('dd-trace/ext/tags')

const span = tracer.startSpan('web.request')

// Always drop the trace
span.setTag(tags.MANUAL_DROP)
//method impl follows

using Datadog.Trace;

using(var scope = Tracer.Instance.StartActive("my-operation"))
{
    var span = scope.Span;

    // Always keep this trace
    span.SetTag(Datadog.Trace.Tags.ManualKeep, "true");
    //method impl follows
}

using Datadog.Trace;

using(var scope = Tracer.Instance.StartActive("my-operation"))
{
    var span = scope.Span;

    // Always drop this trace
    span.SetTag(Datadog.Trace.Tags.ManualDrop, "true");
    //method impl follows
}

<?php
  $tracer = \DDTrace\GlobalTracer::get();
  $span = $tracer->getActiveSpan();

  if (null !== $span) {
    // Always keep this trace
    $span->setTag(\DDTrace\Tag::MANUAL_KEEP, true);
  }
?>

<?php
  $tracer = \DDTrace\GlobalTracer::get();
  $span = $tracer->getActiveSpan();

  if (null !== $span) {
    // Always drop this trace
    $span->setTag(\DDTrace\Tag::MANUAL_DROP, true);
  }
?>

...
#include <datadog/tags.h>
#include <datadog/trace_segment.h>
#include <datadog/sampling_priority.h>
...

dd::SpanConfig span_cfg;
span_cfg.resource = "operation_name";

auto span = tracer.create_span(span_cfg);
// Always keep this trace
span.trace_segment().override_sampling_priority(int(dd::SamplingPriority::USER_KEEP));
//method impl follows

...
#include <datadog/tags.h>
#include <datadog/trace_segment.h>
#include <datadog/sampling_priority.h>
...

using namespace dd = datadog::tracing;

dd::SpanConfig span_cfg;
span_cfg.resource = "operation_name";

auto another_span = tracer.create_span(span_cfg);
// Always drop this trace
span.trace_segment().override_sampling_priority(int(dd::SamplingPriority::USER_DROP));
//method impl follows

Set Manual Keep before context propagation. If set after context propagation, the entire trace may not be kept across services. Because this decision is set at the tracing client, the trace can still be dropped by the Agent or server based on sampling rules.

Single spans

ingestion_reason: single_span

If you need to sample a specific span but don’t need the full trace, SDKs allow you to set a sampling rate for a single span.

For example, if you are building metrics from spans to monitor specific services, you can configure span sampling rules so that these metrics are based on 100% of application traffic, without ingesting 100% of traces for all requests flowing through the service.

This feature is available for Datadog Agent v7.40.0+.

Note: Single span sampling rules cannot be used to drop spans that are kept by head-based sampling, only to keep additional spans that are dropped by head-based sampling.

@env DD_SPAN_SAMPLING_RULES=[{"service": "my-service", "name": "http.request", "sample_rate":1.0, "max_per_second": 50}]

@env DD_SPAN_SAMPLING_RULES=[{"service": "my-service", "name": "http.request", "sample_rate":1.0, "max_per_second": 50}]

@env DD_SPAN_SAMPLING_RULES=[{"service": "my-service", "name": "http.request", "sample_rate":1.0, "max_per_second": 50}]

@env DD_SPAN_SAMPLING_RULES=[{"service": "my-service", "name": "http.request", "sample_rate":1.0, "max_per_second": 50}]

@env DD_SPAN_SAMPLING_RULES=[{"resource": "POST /api/create_issue", "tags": { "priority":"high" }, "sample_rate":1.0}]

@env DD_SPAN_SAMPLING_RULES=[{"service": "my-service", "name": "http.request", "sample_rate":1.0, "max_per_second": 50}]

@env DD_SPAN_SAMPLING_RULES=[{"service": "my-service", "name": "http.request", "sample_rate":1.0, "max_per_second": 50}]

@env DD_SPAN_SAMPLING_RULES=[{"service": "my-service", "name": "http.request", "sample_rate":1.0, "max_per_second": 50}]

#using powershell
$env:DD_SPAN_SAMPLING_RULES='[{"service": "my-service", "name": "http.request", "sample_rate":1.0, "max_per_second": 50}]'

#using JSON file   
{
    "DD_SPAN_SAMPLING_RULES": "[{\"service\": \"my-service\", \"name\": \"http.request\", \"sample_rate\": 1.0, \"max_per_second\": 50}]"
}

Product ingested spans

RUM traces

ingestion_reason:rum

A request from a web or mobile application generates a trace when the backend services are instrumented. The APM integration with Real User Monitoring links web and mobile application requests to their corresponding backend traces so you can see your full frontend and backend data through one lens.

Beginning with version 4.30.0 of the RUM browser SDK, you can control ingested volumes and keep a sampling of the backend traces by configuring the traceSampleRate initialization parameter. Set traceSampleRate to a number between 0 and 100. If no traceSampleRate value is set, a default of 100% of the traces coming from the browser requests are sent to Datadog.

You can also control the trace sampling rate in other SDKs:

Synthetic traces

ingestion_reason:synthetics and ingestion_reason:synthetics-browser

HTTP and browser tests generate traces when the backend services are instrumented. The APM integration with Synthetic Testing links your synthetic tests with the corresponding backend traces. Navigate from a test run that failed to the root cause of the issue by looking at the trace generated by that test run.

By default, 100% of synthetic HTTP and browser tests generate backend traces.

Other products

Some additional ingestion reasons are attributed to spans that are generated by specific Datadog products:

Ingestion mechanisms in OpenTelemetry

ingestion_reason:otel

Depending on your setup with the OpenTelemetry SDKs (using the OpenTelemetry Collector or the Datadog Agent), you have multiple ways of controlling ingestion sampling. See Ingestion Sampling with OpenTelemetry for details about the options available for sampling at the OpenTelemetry SDK, OpenTelemetry Collector, and Datadog Agent level in various OpenTelemetry setups.

Further Reading

Additional helpful documentation, links, and articles:

Ingestion ControlsDOCUMENTATION Trace RetentionDOCUMENTATION Usage MetricsDOCUMENTATION Optimizing Datadog at scale: Cost-efficient observability at ZendeskBLOG APM Rate Limiting and RetentionLEARNING CENTER