[martincostello]

Reduce hub invocation activity overhead

Reduce the overhead of activity for SignalR hub invocations.

Description

• Pass hub activity tags to ActivitySource in a reusable thread-static list to avoid boxing.
• Cache the activity display name per hub method.
• Cache the link to the connection's original activity per connection.
• Avoid performance regression when using TagList.

Benchmarks

DefaultHubDispatcherTelemetryBenchmark.Invocation

Scenario	Baseline time	Candidate time	Time	Baseline allocated	Candidate allocated	Allocated
No telemetry	172.0 ns	165.9 ns	−3.5%	552 B	552 B	0%
Traces	751.7 ns	502.8 ns	−33.1%	2,064 B	1,488 B	−27.9%

Traced invocations on main also pay a JIT AVX to SSE transition penalty that the branch avoids (dotnet/runtime#133784). With DOTNET_EnableAVX=0, a traced invocation takes 539.9 ns on main and 492.2 ns on the branch.

Detailed results

main

BenchmarkDotNet=v0.13.0, OS=Windows 10.0.26200
Unknown processor
.NET SDK=11.0.100-rc.1.26420.103
  [Host] : .NET 11.0.0 (11.0.26.45408), X64 RyuJIT

Affinity=00000000111111111111  Toolchain=InProcessEmitToolchain  RunStrategy=Throughput

Method	Tracing	Mean	Error	StdDev	Op/s	Gen 0	Gen 1	Gen 2	Allocated
Invocation	False	172.0 ns	2.61 ns	2.31 ns	5,814,777.3	0.0439	-	-	552 B
Invocation	True	764.6 ns	13.49 ns	12.62 ns	1,307,925.8	0.1640	-	-	2,064 B

PR

BenchmarkDotNet=v0.13.0, OS=Windows 10.0.26200
Unknown processor
.NET SDK=11.0.100-rc.1.26420.103
  [Host] : .NET 11.0.0 (11.0.26.45408), X64 RyuJIT

Affinity=00000000111111111111  Toolchain=InProcessEmitToolchain  RunStrategy=Throughput

Method	Tracing	Mean	Error	StdDev	Median	Op/s	Gen 0	Gen 1	Gen 2	Allocated
Invocation	False	169.1 ns	2.94 ns	4.03 ns	169.3 ns	5,912,187.4	0.0439	-	-	552 B
Invocation	True	507.3 ns	9.74 ns	27.14 ns	493.9 ns	1,971,107.7	0.1183	-	-	1,488 B

[dotnet-policy-service]

Thanks for your PR, @martincostello. Someone from the team will get assigned to your PR shortly and we'll get it reviewed.

[Copilot]

Copilot review overview

🔵 Needs a closer look

Low-level thread-static reuse and concurrent telemetry caching warrant final human review.

Review effort: Balanced
Findings: None

What changed in this PR

Optimizes SignalR hub invocation telemetry to reduce traced-request allocations and execution time.

Changes:

• Reuses tag collections and cached activity links/display names.
• Generalizes endpoint tagging for TagList and list-backed collections.
• Adds activity behavior tests and a telemetry benchmark.

File	Description
HubConnectionHandlerTests.Activity.cs	Tests resolved activity method names.
HubMethodDescriptor.cs	Caches activity display names.
DefaultHubDispatcher.cs	Reuses tags and cached telemetry metadata.
HubConnectionContext.cs	Caches the original activity link.
DefaultHubDispatcherTelemetryBenchmark.cs	Benchmarks invocation telemetry overhead.
ConnectionEndpointTags.cs	Supports generic tag collections without boxing.

---

💡 Add a code-review agent skill for context-aware, tailored reviews. Learn more in the docs.

[dotnet-policy-service]

Looks like this PR hasn't been active for some time and the codebase could have been changed in the meantime.
To make sure no conflicting changes have occurred, please rerun validation before merging. You can do this by leaving an /azp run comment here (requires commit rights), or by simply closing and reopening.

[dotnet-policy-service]

Looks like this PR hasn't been active for some time and the codebase could have been changed in the meantime.
To make sure no conflicting changes have occurred, please rerun validation before merging. You can do this by leaving an /azp run comment here (requires commit rights), or by simply closing and reopening.

[dotnet-policy-service]

Looks like this PR hasn't been active for some time and the codebase could have been changed in the meantime.
To make sure no conflicting changes have occurred, please rerun validation before merging. You can do this by leaving an /azp run comment here (requires commit rights), or by simply closing and reopening.