Perf optimization for protobuf instrumentation #6694
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Execution-Time Benchmarks Report ⏱️Execution-time results for samples comparing the following branches/commits: Execution-time benchmarks measure the whole time it takes to execute a program. And are intended to measure the one-off costs. Cases where the execution time results for the PR are worse than latest master results are shown in red. The following thresholds were used for comparing the execution times:
Note that these results are based on a single point-in-time result for each branch. For full results, see the dashboard. Graphs show the p99 interval based on the mean and StdDev of the test run, as well as the mean value of the run (shown as a diamond below the graph). gantt
title Execution time (ms) FakeDbCommand (.NET Framework 4.6.2)
dateFormat X
axisFormat %s
todayMarker off
section Baseline
This PR (6694) - mean (69ms) : 66, 72
. : milestone, 69,
master - mean (69ms) : 66, 72
. : milestone, 69,
section CallTarget+Inlining+NGEN
This PR (6694) - mean (1,002ms) : 980, 1024
. : milestone, 1002,
master - mean (998ms) : 973, 1022
. : milestone, 998,
gantt
title Execution time (ms) FakeDbCommand (.NET Core 3.1)
dateFormat X
axisFormat %s
todayMarker off
section Baseline
This PR (6694) - mean (103ms) : 101, 105
. : milestone, 103,
master - mean (102ms) : 99, 105
. : milestone, 102,
section CallTarget+Inlining+NGEN
This PR (6694) - mean (674ms) : 659, 690
. : milestone, 674,
master - mean (678ms) : 659, 697
. : milestone, 678,
gantt
title Execution time (ms) FakeDbCommand (.NET 6)
dateFormat X
axisFormat %s
todayMarker off
section Baseline
This PR (6694) - mean (89ms) : 87, 91
. : milestone, 89,
master - mean (89ms) : 87, 90
. : milestone, 89,
section CallTarget+Inlining+NGEN
This PR (6694) - mean (627ms) : 607, 647
. : milestone, 627,
master - mean (631ms) : 615, 647
. : milestone, 631,
gantt
title Execution time (ms) HttpMessageHandler (.NET Framework 4.6.2)
dateFormat X
axisFormat %s
todayMarker off
section Baseline
This PR (6694) - mean (191ms) : 188, 195
. : milestone, 191,
master - mean (191ms) : 187, 196
. : milestone, 191,
section CallTarget+Inlining+NGEN
This PR (6694) - mean (1,110ms) : 1086, 1135
. : milestone, 1110,
master - mean (1,114ms) : 1079, 1149
. : milestone, 1114,
gantt
title Execution time (ms) HttpMessageHandler (.NET Core 3.1)
dateFormat X
axisFormat %s
todayMarker off
section Baseline
This PR (6694) - mean (272ms) : 267, 277
. : milestone, 272,
master - mean (271ms) : 267, 276
. : milestone, 271,
section CallTarget+Inlining+NGEN
This PR (6694) - mean (865ms) : 838, 892
. : milestone, 865,
master - mean (869ms) : 838, 901
. : milestone, 869,
gantt
title Execution time (ms) HttpMessageHandler (.NET 6)
dateFormat X
axisFormat %s
todayMarker off
section Baseline
This PR (6694) - mean (263ms) : 258, 269
. : milestone, 263,
master - mean (262ms) : 258, 266
. : milestone, 262,
section CallTarget+Inlining+NGEN
This PR (6694) - mean (848ms) : 810, 886
. : milestone, 848,
master - mean (848ms) : 816, 880
. : milestone, 848,
|
Benchmarks Report for tracer 🐌Benchmarks for #6694 compared to master:
The following thresholds were used for comparing the benchmark speeds:
Allocation changes below 0.5% are ignored. Benchmark detailsBenchmarks.Trace.ActivityBenchmark - Same speed ✔️ Same allocations ✔️Raw results
Benchmarks.Trace.AgentWriterBenchmark - Same speed ✔️ Same allocations ✔️Raw results
Benchmarks.Trace.AspNetCoreBenchmark - Same speed ✔️ Same allocations ✔️Raw results
Benchmarks.Trace.CIVisibilityProtocolWriterBenchmark - Same speed ✔️ Same allocations ✔️Raw results
Benchmarks.Trace.DbCommandBenchmark - Same speed ✔️ Same allocations ✔️Raw results
Benchmarks.Trace.ElasticsearchBenchmark - Same speed ✔️ Same allocations ✔️Raw results
Benchmarks.Trace.GraphQLBenchmark - Same speed ✔️ Same allocations ✔️Raw results
Benchmarks.Trace.HttpClientBenchmark - Same speed ✔️ Same allocations ✔️Raw results
Benchmarks.Trace.ILoggerBenchmark - Same speed ✔️ Same allocations ✔️Raw results
Benchmarks.Trace.Log4netBenchmark - Same speed ✔️ Same allocations ✔️Raw results
Benchmarks.Trace.NLogBenchmark - Same speed ✔️ Same allocations ✔️Raw results
Benchmarks.Trace.RedisBenchmark - Same speed ✔️ Same allocations ✔️Raw results
Benchmarks.Trace.SerilogBenchmark - Same speed ✔️ Same allocations ✔️Raw results
Benchmarks.Trace.SpanBenchmark - Slower
|
| Benchmark | diff/base | Base Median (ns) | Diff Median (ns) | Modality |
|---|---|---|---|---|
| Benchmarks.Trace.SpanBenchmark.StartFinishSpan‑net472 | 1.149 | 591.85 | 680.08 |
Raw results
| Branch | Method | Toolchain | Mean | StdError | StdDev | Gen 0 | Gen 1 | Gen 2 | Allocated |
|---|---|---|---|---|---|---|---|---|---|
| master | StartFinishSpan |
net6.0 | 447ns | 0.887ns | 3.43ns | 0.00806 | 0 | 0 | 576 B |
| master | StartFinishSpan |
netcoreapp3.1 | 600ns | 0.831ns | 3ns | 0.00782 | 0 | 0 | 576 B |
| master | StartFinishSpan |
net472 | 590ns | 1.96ns | 7.33ns | 0.0915 | 0 | 0 | 578 B |
| master | StartFinishScope |
net6.0 | 482ns | 0.943ns | 3.53ns | 0.0097 | 0 | 0 | 696 B |
| master | StartFinishScope |
netcoreapp3.1 | 717ns | 1.98ns | 7.68ns | 0.00962 | 0 | 0 | 696 B |
| master | StartFinishScope |
net472 | 886ns | 1.84ns | 7.12ns | 0.105 | 0 | 0 | 658 B |
| #6694 | StartFinishSpan |
net6.0 | 465ns | 0.633ns | 2.28ns | 0.00806 | 0 | 0 | 576 B |
| #6694 | StartFinishSpan |
netcoreapp3.1 | 565ns | 0.597ns | 2.23ns | 0.0077 | 0 | 0 | 576 B |
| #6694 | StartFinishSpan |
net472 | 679ns | 0.809ns | 3.03ns | 0.0916 | 0 | 0 | 578 B |
| #6694 | StartFinishScope |
net6.0 | 482ns | 0.83ns | 3.11ns | 0.00977 | 0 | 0 | 696 B |
| #6694 | StartFinishScope |
netcoreapp3.1 | 752ns | 0.856ns | 3.2ns | 0.00929 | 0 | 0 | 696 B |
| #6694 | StartFinishScope |
net472 | 860ns | 1.69ns | 6.56ns | 0.104 | 0 | 0 | 658 B |
Benchmarks.Trace.TraceAnnotationsBenchmark - Same speed ✔️ Same allocations ✔️
Raw results
| Branch | Method | Toolchain | Mean | StdError | StdDev | Gen 0 | Gen 1 | Gen 2 | Allocated |
|---|---|---|---|---|---|---|---|---|---|
| master | RunOnMethodBegin |
net6.0 | 679ns | 0.696ns | 2.69ns | 0.00986 | 0 | 0 | 696 B |
| master | RunOnMethodBegin |
netcoreapp3.1 | 882ns | 0.557ns | 2.01ns | 0.0092 | 0 | 0 | 696 B |
| master | RunOnMethodBegin |
net472 | 1.09μs | 1.9ns | 7.36ns | 0.104 | 0 | 0 | 658 B |
| #6694 | RunOnMethodBegin |
net6.0 | 683ns | 0.775ns | 3ns | 0.00979 | 0 | 0 | 696 B |
| #6694 | RunOnMethodBegin |
netcoreapp3.1 | 926ns | 1.81ns | 7.02ns | 0.00924 | 0 | 0 | 696 B |
| #6694 | RunOnMethodBegin |
net472 | 1μs | 1.84ns | 7.12ns | 0.104 | 0 | 0 | 658 B |
lucaspimentel
added
type:performance
tracer/src/Datadog.Trace/ClrProfiler/AutoInstrumentation/Protobuf/Helper.cs
Outdated
Show resolved
Hide resolved
lucaspimentel
approved these changes
Feb 24, 2025
Co-authored-by: Lucas Pimentel <[email protected]>
vandonr
force-pushed
the
vandonr/proto
branch
from
3ee76fd to
345fb91
Compare
bouwkast
approved these changes
Feb 25, 2025
Comment on lines
+32
to
+35
| // For performance reasons, we want to do the actual instrumentation work with a Duck constraint, | ||
| // but to be able to disable the instrumentation we need the raw type | ||
| // so we use 2 different methods to have access to both when we need it. | ||
| // Note: Disabling OnMethodBegin means the OnMethodEnd will not be called afterward. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary of changes
Stop relying on the proto source file to know if we're dealing with an internal protobuf message.
Instead, we check the type at the very beginning, and disable the instrumentation for that type if we detect it's a google type.
This means we can also get rid of all the dark magic we had around making sure we can access the descriptor, because we now know we'll only do it for regular messages.
Reason for change
it's better perf
Implementation details
I use an other
OnMethodXxxmethod to be able to access the raw type, I explained it in the comments.See this thread for more.
Test coverage
added a path in the integration tests that checks that
Other details