flowcontrol: add benchmark suite

[LukeAVanDrie]

What type of PR is this?

/kind test

What this PR does / why we need it:

This PR introduces a comprehensive, steady-state benchmarking suite for the Flow Control layer. This suite is essential for establishing baseline performance metrics, identifying concurrency bottlenecks, and ensuring we prevent regressions as the system scales.

Results:

Results were collected on Intel(R) Xeon(R) CPU @ 2.20GHz using 8 cores for reference via:

go test -v -bench=. -run="^$" \            
  -benchtime=1s \
  -count=5 \
  -cpuprofile=baseline_cpu.prof \
  -memprofile=baseline_mem.prof \
  -blockprofile=baseline_block.prof \
  -mutexprofile=baseline_mutex.prof \
  ./pkg/epp/flowcontrol/benchmark/... | tee baseline.txt

The bulk of the testing is driven by BenchmarkFlowController_PerformanceMatrix which affixes and sweeps the following values, yielding a full performance hypercube:

• Egress concurrency (modeling IFR-based saturation): {free-flow (always unsaturated), 1, 100}
• # Shards (data parallelism): {1, 8}
• # Priorities: {1, 8}
• # Flows: {10, 50000}
• Ingress concurrency (# concurrent requests): {10, 5000, 50000}

In short, the Flow Control layer is working efficiently. The dominant blocking factor is standard channel backpressure (selectgo) which is expected with our queuing strategy. There is minimal, localized lock contention. The primary limiter of raw macroscopic throughput is Garbage Collection overhead generated by allocations during IterateQueues and vending the FlowQueueAcessor. Both of these are clear, isolated targets for zero-allocation strategies. I have already fixed and validated this locally. I will send targeted performance improvements as followup PRs.

• benchstat
• tops.txt

Which issue(s) this PR fixes:
Part of #2087

Does this PR introduce a user-facing change?:

NONE

[k8s-ci-robot]

@LukeAVanDrie: The label(s) kind/test cannot be applied, because the repository doesn't have them.

Details

In response to this:

What type of PR is this?

/kind test

What this PR does / why we need it:

This PR introduces a comprehensive, steady-state benchmarking suite for the Flow Control layer. This suite is essential for establishing baseline performance metrics, identifying concurrency bottlenecks, and ensuring we prevent regressions as the system scales.

Which issue(s) this PR fixes:
Part of #2087

Does this PR introduce a user-facing change?:

NONE

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository.

[netlify]

✅ Deploy Preview for gateway-api-inference-extension ready!

Name	Link
🔨 Latest commit	a8b6b2d
🔍 Latest deploy log	https://app.netlify.com/projects/gateway-api-inference-extension/deploys/69b0a04e9816590008f7d22c
😎 Deploy Preview	https://deploy-preview-2539--gateway-api-inference-extension.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.

---

To edit notification comments on pull requests, go to your Netlify project configuration.

[LukeAVanDrie]

/assign @kfswain

[ahg-g]

Great, unless I missed it, can you pls describe the details of the workload (request rate, number of flows, number of priorities) you benchmarked in the PR description?

[LukeAVanDrie]

Great, unless I missed it, can you pls describe the details of the workload (request rate, number of flows, number of priorities) you benchmarked in the PR description?

This is defined in benchmark_test.go and well-documented there. I run the full matrix of these values (affixing all and sweeping one), yielding the full performance hypercube. Full results are in the benchstat file linked in the PR description. Pprof data (tops and charts) is amortized across the full testing matrix, not for a specific configuration.

I just updated the PR description with this information.

[ahg-g]

/lgtm
/approve

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: ahg-g, LukeAVanDrie

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Details

Needs approval from an approver in each of these files:

• OWNERS [ahg-g]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[ahg-g]

/lgtm