Improve OpenApiWalker performance

[Martin Costello (martincostello)]

Benchmark results: #2459 (comment)

Given that the hardware between my local setup and GitHub Actions is quite different, I suggest merging this PR once happy with the code changes with the failing benchmark comparison, then downloading the results from the run in main and opening a new PR to use those files to replace the contents of performance/benchmark/BenchmarkDotNet.Artifacts/results and update the baseline.

Changes

• Avoid allocations from lambda closures.
• Avoid allocations from context strings.
• Avoid allocations from copying arrays.
• Remove redundant null checks.
• Ignore BenchmarkDotNet profiler files.
• Ignore Visual Studio profiler session files.
• Publish benchmark results to GitHub Actions workflow artifacts.

[Copilot]

Pull Request Overview

This PR optimizes the performance of the OpenApiWalker class by eliminating allocations from lambda closures, string concatenations, and array copying operations. The changes focus on replacing the existing Walk(context, action) pattern with more efficient WalkItem and WalkDictionary helper methods that avoid capturing variables in closures.

Key changes include:

• Replace lambda closures with direct method calls to avoid allocation overhead
• Add specialized WalkItem and WalkDictionary helper methods for efficient traversal
• Implement optimized iteration patterns for collections to avoid array copying
• Publish benchmark results as GitHub Actions artifacts for performance tracking

Reviewed Changes

Copilot reviewed 2 out of 3 changed files in this pull request and generated 4 comments.

File	Description
src/Microsoft.OpenApi/Services/OpenApiWalker.cs	Core performance optimizations replacing lambda-based traversal with direct method calls
.github/workflows/ci-cd.yml	Add artifact publishing for benchmark results

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[Martin Costello (martincostello)]

FYI I ran each of the benchmarks using the Visual Studio profiler to get the allocations, and these are the top types:

Even more refactoring of the walker in a future PR to remove the delegates would probably significantly reduce the allocations:

[Martin Costello (martincostello)]

Some of these might be new from the refactoring actually, so I'll see if I can shave it down more here...

[Martin Costello (martincostello)]

With the latest changes:

Method	Mean Before	Mean After	Ratio	Memory Before	Memory After	Ratio
PetStoreYaml	265.6 μs	293.3 μs	1.10	387.12 KB	386.91 KB	0.99
PetStoreJson	106.6 μs	134.4 μs	1.26	249.26 KB	249.06 KB	0.99
GHESYaml	774,833.9 μs	788,176.8 μs	1.01	400088.73 KB	384130.43 KB	0.96
GHESJson	364,114.2 μs	368,179.1 μs	1.01	261558.87 KB	262008.28 KB	1.00

BenchmarkDotNet v0.15.2, Windows 11 (10.0.26100.4946/24H2/2024Update/HudsonValley)
13th Gen Intel Core i7-13700H 2.90GHz, 1 CPU, 20 logical and 14 physical cores
.NET SDK 8.0.413
[Host] : .NET 8.0.19 (8.0.1925.36514), X64 RyuJIT AVX2
ShortRun : .NET 8.0.19 (8.0.1925.36514), X64 RyuJIT AVX2
Job=ShortRun IterationCount=3 LaunchCount=1 WarmupCount=3 

Method	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
PetStoreYaml	293.3 μs	23.48 μs	1.29 μs	31.2500	5.8594	-	386.91 KB
PetStoreJson	134.4 μs	49.91 μs	2.74 μs	19.5313	4.8828	-	249.06 KB
GHESYaml	788,176.8 μs	246,902.17 μs	13,533.54 μs	35000.0000	19000.0000	4000.0000	384130.43 KB
GHESJson	368,179.1 μs	23,852.48 μs	1,307.43 μs	20000.0000	11000.0000	2000.0000	262008.28 KB

[Martin Costello (martincostello)]

Action<T> has dropped off top of the allocations now:

[Vincent Biret (baywet)]

This is some great refactoring that was overdue here! Thanks for taking the time. I have left a few comments.

And yes... it's not the first time I see the action context capture using a lot of memory at scale...

[Martin Costello (martincostello)]

Here's all the relevant benchmark data taken from the CI logs.

TL;DR:

• Allocations are now less than before Validate OpenAPI schema references #2459 was merged
• Runtime performance is slightly worse for the PetStore document, but is faster for GHES.

10b46b9

Method	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
PetStoreYaml	502.9 us	85.92 us	4.71 us	23.4375	3.9063	-	387.12 KB
PetStoreJson	215.0 us	10.97 us	0.60 us	14.6484	3.9063	-	248.97 KB
GHESYaml	1,063,967.0 us	51,235.65 us	2,808.40 us	27000.0000	20000.0000	3000.0000	400086.95 KB
GHESJson	509,285.0 us	133,784.60 us	7,333.19 us	17000.0000	10000.0000	2000.0000	261557.86 KB

90b3966

Method	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
PetStoreYaml	553.4 us	19.28 us	1.06 us	26.3672	5.8594	-	445.71 KB
PetStoreJson	276.7 us	55.86 us	3.06 us	17.5781	3.9063	-	307.56 KB
GHESYaml	1,055,962.2 us	291,474.86 us	15,976.72 us	28000.0000	20000.0000	3000.0000	422172.41 KB
GHESJson	509,499.6 us	119,766.91 us	6,564.83 us	18000.0000	10000.0000	2000.0000	283643.26 KB

This PR

Method	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
PetStoreYaml	511.4 us	68.15 us	3.74 us	23.4375	3.9063	-	386.9 KB
PetStoreJson	233.9 us	20.54 us	1.13 us	13.6719	3.9063	-	248.75 KB
GHESYaml	1,063,226.7 us	317,042.02 us	17,378.14 us	26000.0000	20000.0000	3000.0000	384128.47 KB
GHESJson	472,314.9 us	54,982.85 us	3,013.80 us	16000.0000	9000.0000	2000.0000	245598.41 KB

[Martin Costello (martincostello)]

With the factoring suggestions the numbers get slightly worse than without, but it's probably within the realms of noise.

Method	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
PetStoreYaml	533.9 us	467.5 us	25.62 us	23.4375	3.9063	-	387.26 KB
PetStoreJson	243.2 us	251.9 us	13.81 us	13.6719	1.9531	-	249.1 KB
GHESYaml	1,044,990.0 us	80,027.6 us	4,386.59 us	26000.0000	20000.0000	3000.0000	384490.94 KB
GHESJson	481,887.0 us	222,646.4 us	12,204.00 us	16000.0000	9000.0000	2000.0000	245957.72 KB

[Martin Costello (martincostello)]

Vincent Biret (@baywet) Assuming you're otherwise happy with the PR and the changes to the perf numbers, do you want me to take the files from the latest CI run on this PR and check them in as the updated baseline (which should then make the PR green on the next run)?

[Vincent Biret (baywet)]

Thank you for making the changes!

[Martin Costello (martincostello)]

Latest numbers:

Method	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
PetStoreYaml	529.5 us	62.50 us	3.43 us	23.4375	3.9063	-	387.26 KB
PetStoreJson	240.8 us	15.69 us	0.86 us	13.6719	1.9531	-	249.1 KB
GHESYaml	1,097,576.6 us	100,584.42 us	5,513.37 us	26000.0000	20000.0000	3000.0000	384492.38 KB
GHESJson	516,328.2 us	87,964.22 us	4,821.62 us	16000.0000	9000.0000	2000.0000	245957.5 KB

Would you like me to commit the results from that build into this PR before merge?

[Vincent Biret (baywet)]

Martin Costello (@martincostello) yes please go ahead!