[SDK] Tracer::StartSpan benchmark and optimizations by dbarker · Pull Request #4248 · open-telemetry/opentelemetry-cpp

dbarker · 2026-07-15T01:32:20Z

Fixes # (issue)
The Tracer::StartSpan function currently performs some unnecessary dynamic memory allocation (allocating a SpanContext and potentially a DefaultSpan) and copying of global static shared pointers by constructing temporary span contexts (which call TraceState::GetDefault()).

This latency can be removed without breaking changes. The added benchmark showed a latency reduction of ~11-23% for the range of span creation scenarios (tracer disabled, span with no parent, implicit parent, and explicit parent).

Changes

Adds a benchmark for Tracer::StartSpan. See baseline vs optimized benchmark results
Implements several optimizations to reduce dynamic memory allocation and shared_ptr copies (9746077)
- Use the DefaultSpan as a non-recording span with a stack allocated SpanContext
- Allocate one non recording span per tracer on construction to serve as the fallback noop span
- Updates the SDK span to take a stack allocated SpanContext and the SamplingResult
- Set the SamplingResult attributes in the SDK Span constructor to avoid locking the span mutex unecessarily
- Changes the default value of the SpanStartOptions::parent to an empty Context instead of the more expensive SpanContext.
- use std::make_shared to allocate the span when building with exceptions
- clean up StartSpan to not instantiate and copy SpanContext if not needed.
- add GetSpanContext helper to avoid dynamically allocating a DefaultSpan object to get an invalid SpanContext (trace::GetSpan is no longer called).
Update tracer test cases to check for a non-recording span using the Span::IsRecording method

For significant contributions please make sure you have completed the following items:

CHANGELOG.md updated for non-trivial changes
Unit tests have been added
Changes in public API reviewed

…n span start options changed

codecov · 2026-07-15T02:50:04Z

Codecov Report

❌ Patch coverage is 89.18919% with 8 lines in your changes missing coverage. Please review.
✅ Project coverage is 81.12%. Comparing base (54f57f3) to head (b208179).

Files with missing lines	Patch %	Lines
sdk/src/trace/tracer.cc	87.10%	8 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #4248      +/-   ##
==========================================
- Coverage   81.13%   81.12%   -0.00%     
==========================================
  Files         444      444              
  Lines       18848    18861      +13     
==========================================
+ Hits        15291    15300       +9     
- Misses       3557     3561       +4

Files with missing lines	Coverage Δ
api/include/opentelemetry/trace/context.h	`95.66% <100.00%> (ø)`
api/include/opentelemetry/trace/default_span.h	`36.85% <ø> (+21.06%)`	⬆️
sdk/include/opentelemetry/sdk/trace/sampler.h	`100.00% <100.00%> (ø)`
sdk/include/opentelemetry/sdk/trace/tracer.h	`100.00% <ø> (ø)`
sdk/src/trace/span.cc	`93.75% <100.00%> (+0.18%)`	⬆️
sdk/src/trace/span.h	`100.00% <100.00%> (ø)`
sdk/src/trace/tracer.cc	`80.22% <87.10%> (-7.28%)`	⬇️

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

…te the NonRecordingSpan. Fix iwyu warnings. Cleanup

dbarker · 2026-07-15T13:56:44Z

+
+// clang-format off
+//
+// ~/build/sdk/test/trace/tracer_benchmark --benchmark_repetitions=5 --benchmark_display_aggregates_only=true


The optimizations in this PR result in a 11-23% reduction in latency for StartSpan compared to main.

Benchmark Baseline (mean ± stddev) Optimized (mean ± stddev) Difference %

BM_StartSpanTracerDisabled 6.85 ± 0.025 ns 5.24 ± 0.046 ns -23.5%

BM_StartSpan 187 ± 1.22 ns 152 ± 2.74 ns -18.7%

BM_StartSpanWithScope 255 ± 1.09 ns 218 ± 1.75 ns -14.5%

BM_StartSpanWithImplicitParent 172 ± 0.45 ns 153 ± 0.80 ns -11.0%

BM_StartSpanWithExplicitParentContext 173 ± 0.34 ns 154 ± 0.90 ns -11.0%

BM_StartSpanWithExplicitRootContext 187 ± 0.68 ns 148 ± 0.69 ns -20.9%

dbarker · 2026-07-15T14:02:18Z

+// BM_TraceIdRatioBasedSamplerShouldSample_median       4.00 ns         4.00 ns            5
+// BM_TraceIdRatioBasedSamplerShouldSample_stddev      0.032 ns        0.032 ns            5
+// BM_TraceIdRatioBasedSamplerShouldSample_cv           0.80 %          0.80 %             5
+// BM_SpanCreation_mean                                  205 ns          205 ns            5


Comparison with main and this PR on span sampler benchmark. The NoopSpanCreation case is where the tracer is enabled but the sampler decides to not sample the span (AlwaysOffSampler).

Benchmark Baseline (mean ± stddev) Optimized (mean ± stddev) Difference %

BM_SpanCreation 235 ± 0.62 ns 205 ± 1.00 ns -12.8%

BM_NoopSpanCreation 59.3 ± 0.42 ns 27.7 ± 0.14 ns -53.3%

…for more consistent results

…r call of GetSpanContext by removing the call to HasKey since it gets a context value anyways

… main

dbarker · 2026-07-17T17:42:06Z

 inline SpanContext GetSpanContext(const context::Context &context) noexcept
 {
-  if (!context.HasKey(kSpanKey))
+  const context::ContextValue context_value = context.GetValue(kSpanKey);


removed context.HasKey call since it just calls context.GetValue(key) and checks for the nostd::monostate. The span key lookup and context value copy can be done once.

dbarker · 2026-07-17T17:50:15Z

  // parent of the currently active Span (if any) in the current context.
  //
-  nostd::variant<SpanContext, context::Context> parent = SpanContext::GetInvalid();
+  nostd::variant<SpanContext, context::Context> parent = context::Context{};


For most callers the value of parent is never changed. For callers that set a parent the initial value is often default constructed then replaced. To keep this change ABI safe we don't add a nostd::monostate (which is really what we want) and instead construct the cheapest object by default. The SpanContext is a complex object with a shared_ptr<TraceState> that is initialized to a global TraceState default object. The full Context is simpler and only has a shared_ptr<DataList> member that is default initialized to null.

…ibutes from the sampler. cleanup

dbarker added 7 commits July 14, 2026 20:35

Add tracer benchmark and record baseline

27c18fd

Performance optimizations for SDK Tracer::StartSpan

9746077

fix iwyu warnings and cleanup

e472646

fix clang-tidy warnings and clean up

2192679

fix noexcept build

f3f6d01

add missing file

84cab3c

fix opentracing shim test after default value of the parent context i…

9caf280

…n span start options changed

The DefaultSpan is a non-recording span, use it in tracer.cc and dele…

b0140b2

…te the NonRecordingSpan. Fix iwyu warnings. Cleanup

dbarker commented Jul 15, 2026

View reviewed changes

Set the in memory span exporter buffer to 0 for the tracer benchmark …

38a0f60

…for more consistent results

dbarker marked this pull request as ready for review July 15, 2026 16:52

dbarker requested a review from a team as a code owner July 15, 2026 16:52

dbarker and others added 7 commits July 15, 2026 12:55

add changelog entry

4bde64c

Merge branch 'main' into perf_start_span_optimizations

f0e6367

Merge branch 'main' into perf_start_span_optimizations

a815cbd

Merge branch 'main' into perf_start_span_optimizations

4ae94f0

use the api GetSpanContext in StartSpan. Get only one ContextValue pe…

067c2b3

…r call of GetSpanContext by removing the call to HasKey since it gets a context value anyways

update results for the tracer and sampler benchmarks after merging in…

ddb176b

… main

fix iwyu warning

2852840

dbarker commented Jul 17, 2026

View reviewed changes

proost reviewed Jul 18, 2026

View reviewed changes

Comment thread sdk/src/trace/tracer.cc Outdated

dbarker added 5 commits July 18, 2026 10:48

add sampler benchmark to cover samplers that add attributes to the span

3c6e1d0

pass sampling result to the SDK Span constructor in order to add attr…

2e42634

…ibutes from the sampler. cleanup

Merge branch 'main' into perf_start_span_optimizations

251128e

fix iwyu warnings

02d6394

fix bazel.noexcept build

b208179

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[SDK] Tracer::StartSpan benchmark and optimizations#4248

[SDK] Tracer::StartSpan benchmark and optimizations#4248
dbarker wants to merge 21 commits into
open-telemetry:mainfrom
dbarker:perf_start_span_optimizations

dbarker commented Jul 15, 2026 •

edited

Loading

Uh oh!

codecov Bot commented Jul 15, 2026 •

edited

Loading

Uh oh!

dbarker Jul 15, 2026

Uh oh!

dbarker Jul 15, 2026

Uh oh!

dbarker Jul 17, 2026

Uh oh!

dbarker Jul 17, 2026 •

edited

Loading

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Benchmark	Baseline (mean ± stddev)	Optimized (mean ± stddev)	Difference %
`BM_StartSpanTracerDisabled`	6.85 ± 0.025 ns	5.24 ± 0.046 ns	-23.5%
`BM_StartSpan`	187 ± 1.22 ns	152 ± 2.74 ns	-18.7%
`BM_StartSpanWithScope`	255 ± 1.09 ns	218 ± 1.75 ns	-14.5%
`BM_StartSpanWithImplicitParent`	172 ± 0.45 ns	153 ± 0.80 ns	-11.0%
`BM_StartSpanWithExplicitParentContext`	173 ± 0.34 ns	154 ± 0.90 ns	-11.0%
`BM_StartSpanWithExplicitRootContext`	187 ± 0.68 ns	148 ± 0.69 ns	-20.9%

Benchmark	Baseline (mean ± stddev)	Optimized (mean ± stddev)	Difference %
`BM_SpanCreation`	235 ± 0.62 ns	205 ± 1.00 ns	-12.8%
`BM_NoopSpanCreation`	59.3 ± 0.42 ns	27.7 ± 0.14 ns	-53.3%

Uh oh!

Conversation

dbarker commented Jul 15, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Changes

Uh oh!

codecov Bot commented Jul 15, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

dbarker Jul 15, 2026

Choose a reason for hiding this comment

Uh oh!

dbarker Jul 15, 2026

Choose a reason for hiding this comment

Uh oh!

dbarker Jul 17, 2026

Choose a reason for hiding this comment

Uh oh!

dbarker Jul 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

dbarker commented Jul 15, 2026 •

edited

Loading

codecov Bot commented Jul 15, 2026 •

edited

Loading

dbarker Jul 17, 2026 •

edited

Loading