阅读中文版本

Build a Benchmark Runner CLI

Benchmark Runner CLI — a complete walkthrough of performance benchmarking, covering benchmark loader, warmup runner, sample collector, and testing.

benchmark_runner_cli/
├── src/index.ts
├── src/benchmark-loader.ts
├── src/warmup-runner.ts
├── tests/benchmark_runner_cli.test.ts
└── package.json

// Benchmark loader: performance benchmarking core
export function benchmarkloader(input: BenchmarkloaderInput): BenchmarkloaderResult {
  // Build the benchmark loader that anchors performance benchmarking and defines the core data shape.
  return processBenchmarkloader(input);
}

// Warmup runner — handles performance benchmarking Add the warmup runner so
export class Warmuprunner {
  constructor(private opts: WarmuprunnerOptions) {}
  async run(ctx: Context): Promise<WarmuprunnerOutput> {
    return this.execute(ctx);
  }
}

// Sample collector: performance benchmarking stage
export const samplecollector = (next: Next) => async (req: Request) => {
  const out = await transform(req);
  return next(out);
};

// Stat analyzer for performance benchmarking
export async function statanalyzer(stream: AsyncIterable<Chunk>): Promise<Aggregate> {
  let acc = init();
  for await (const c of stream) acc = merge(acc, c);
  return acc;
}

// Regression reporter: performance benchmarking finalization
export function regressionreporter(state: State): Result {
  validate(state);
  return finalize(state);
}

Project overview

This project builds a performance benchmarking system whose reason for existing is running a suite of benchmarks, detecting regressions, and reporting results statistically. The temptation in performance benchmarking is to bolt the interesting part onto a working prototype and discover the hard cases in production; this build inverts that order and treats a noisy environment where variance swamps the regression signal and the reporter flags random fluctuations as the first thing designed, not the last.

You will implement Benchmark loader, Warmup runner, Sample collector, Stat analyzer, and Regression reporter, each as a module with one responsibility and a test that proves it in isolation. The performance benchmarking pipeline is then the wiring that composes them, and the integration test is where the composition is shown to survive a noisy environment where variance swamps the regression signal and the reporter flags random fluctuations.

The walkthrough is written for a developer who has used a performance benchmarking system but has not had to make one honest. It favors explicit data shapes over clever abstractions, and it names the failure each design choice prevents rather than presenting the choice as a style preference.

Each phase ends with something observable — a benchmark loader that returns the right shape, a warmup runner that rejects bad input, an integration test that goes green — so you can stop at any point and still have a performance benchmarking system that works for the cases covered so far.

What you will build

The finished performance benchmarking system exposes five capabilities, each with a focused test:

• Benchmark loader
• Warmup runner
• Sample collector
• Stat analyzer
• Regression reporter

Benchmark loader is the foundation that every later stage assumes; Warmup runner is where performance benchmarking correctness lives; Sample collector handles a noisy environment where variance swamps the regression signal and the reporter flags random fluctuations; Stat analyzer makes the system observable; and Regression reporter is the graceful-degradation layer that keeps overload from becoming corruption.

The build keeps the performance benchmarking logic in ordinary modules you can read in one sitting, and pulls in dependencies only for the parts that are genuinely tedious to hand-roll. That keeps the diff between the tutorial and a real codebase small, and keeps the failure modes visible in code rather than hidden behind a framework.

Architecture and data model

The system is organized around Benchmark loader, Warmup runner, and Sample collector, with Stat analyzer tying them together. Data flows as a set of benchmark functions each run for a sample count after warmup, with median and variance compared to a baseline.

Each component is a module with a pure interface where practical: the same input produces the same output, and side effects are pushed to the edges. That discipline is what makes the phases independently testable, because a test for Sample collector does not need to stand up Benchmark loader and Warmup runner to exercise the interesting branch.

The data model is deliberately explicit. A set of benchmark functions each run for a sample count after warmup, with median and variance compared to a baseline Keeping the shape written down — in a schema, a type, or a fixture — means a change to the shape is a visible diff rather than a silent runtime shift, and it means the integration test can assert against a concrete shape instead of guessing.

Configuration and code are kept apart: the configuration says what performance benchmarking should do, the modules say how. A reviewer can change a threshold or a policy without reading the implementation, which keeps the risky changes small and lets the diff stand as the review.

The failure path through Sample collector is designed before the happy path is finished, because a noisy environment where variance swamps the regression signal and the reporter flags random fluctuations is the case that distinguishes a performance benchmarking system from a demo. Every component declares what it does when its inputs are wrong, and the integration test asserts each of those declarations.

Prerequisites and dependencies

Have these ready before you install anything, so the performance benchmarking setup is not blocked by a missing tool:

• Node.js 20 LTS or later
• Familiarity with performance benchmarking fundamentals
• A local environment to run the test suite

The runtime dependencies are chosen to keep the performance benchmarking datapath legible:

• commander (^12.1.0) — CLI framework
• tinybench (^3.0.0) — Benchmark runtime

# setup for benchmark-runner-cli
pnpm init
pnpm add commander tinybench
pnpm add -D typescript tsx vitest

Versions are pinned to a major range so the performance benchmarking install is reproducible; commit the lockfile. The benchmark loader reads its operational parameters from the environment and validates them at startup, so a missing value fails the boot rather than the first performance benchmarking request.

Directory structure

Keep each performance benchmarking component in its own file so it can be tested without starting the others:

benchmark_runner_cli/
├── src/index.ts
├── src/benchmark-loader.ts
├── src/warmup-runner.ts
├── tests/benchmark_runner_cli.test.ts
└── package.json

Phase 1: Benchmark loader

Build the benchmark loader that anchors performance benchmarking and defines the core data shape.

// Benchmark loader: performance benchmarking core
export function benchmarkloader(input: BenchmarkloaderInput): BenchmarkloaderResult {
  // Build the benchmark loader that anchors performance benchmarking and defines the core data shape.
  return processBenchmarkloader(input);
}

After this phase the benchmark loader works in isolation and is covered by a unit test. The benchmark loader defines the data shape that every later stage assumes, so getting it right first means the later phases compile against a stable contract.

Treat the benchmark loader as the single source of truth for the performance benchmarking data shape. Validate inputs at the boundary and reject early, so a malformed value never reaches Warmup runner or Sample collector where its effect would be harder to localize.

Phase 2: Warmup runner

Add the warmup runner so performance benchmarking can running a suite of correctly.

// Warmup runner — handles performance benchmarking Add the warmup runner so
export class Warmuprunner {
  constructor(private opts: WarmuprunnerOptions) {}
  async run(ctx: Context): Promise<WarmuprunnerOutput> {
    return this.execute(ctx);
  }
}

After this phase the warmup runner works in isolation and is covered by a unit test. This is where performance benchmarking correctness lives: the warmup runner is what makes running a suite of benchmarks, detecting regressions, and reporting results statistically true, and its test is the proof.

Keep the warmup runner pure where you can. A function that takes an input and returns an output is trivial to test and impossible to mock wrong; a class with hidden state is the opposite. Push side effects to the caller and keep the warmup runner about the transformation.

Phase 3: Sample collector

Wire the sample collector to handle a noisy environment where without losing data.

// Sample collector: performance benchmarking stage
export const samplecollector = (next: Next) => async (req: Request) => {
  const out = await transform(req);
  return next(out);
};

After this phase the sample collector works in isolation and is covered by a unit test. The sample collector is the difference between a performance benchmarking system that works on the happy path and one that survives a noisy environment where variance swamps the regression signal and the reporter flags random fluctuations.

Design the failure before the success. Decide what sample collector does when a noisy environment where variance swamps the regression signal and the reporter flags random fluctuations: reject, retry, shed, or degrade — and write that test first. A performance benchmarking system whose failure behavior is unspecified will pick one at random under pressure, and it is usually the wrong one.

Phase 4: Stat analyzer

Integrate the stat analyzer to make performance benchmarking observable and testable end to end.

// Stat analyzer for performance benchmarking
export async function statanalyzer(stream: AsyncIterable<Chunk>): Promise<Aggregate> {
  let acc = init();
  for await (const c of stream) acc = merge(acc, c);
  return acc;
}

After this phase the stat analyzer works in isolation and is covered by a unit test. Observability is not a finishing touch; it is how you know the performance benchmarking system is doing what it claims in production.

Emit one structured event per meaningful action, with enough context to localize a failure without a second round-trip. The stat analyzer is what turns “something is slow” into “the warmup runner is slow for inputs matching this shape,” which is the difference between an alert and an investigation.

Phase 5: Regression reporter

Finalize the regression reporter so the performance benchmarking system degrades gracefully under overload.

// Regression reporter: performance benchmarking finalization
export function regressionreporter(state: State): Result {
  validate(state);
  return finalize(state);
}

After this phase the regression reporter works in isolation and is covered by a unit test. The regression reporter is what keeps overload from becoming corruption: when the performance benchmarking system is asked for more than it can do, it degrades in a defined way instead of failing in an undefined one.

Pick a degradation strategy for regression reporter — shed, queue, or backpressure — and make it explicit in code and in a test. Leaving the behavior to chance is how a performance benchmarking traffic spike becomes a data loss incident.

Testing

Each phase has a focused test: benchmark loader, warmup runner, and sample collector are exercised in isolation, then an integration test drives the full performance benchmarking pipeline and asserts the a noisy environment where case is handled.

Start with unit tests that pin each component’s contract: the benchmark loader returns the documented shape, the warmup runner transforms correctly, and the sample collector fails in the documented way. These tests are fast and they localize regressions to the component that broke.

Then add an integration test that composes the components and drives a realistic performance benchmarking scenario end to end. Assert the happy path first, then deliberately trigger a noisy environment where variance swamps the regression signal and the reporter flags random fluctuations and assert the system degrades rather than corrupts. Because the components are pure where practical, the integration test stays deterministic rather than flaky.

A performance benchmarking suite must also prove the negative: bad input is rejected at the boundary, overload is shed rather than queued indefinitely, and a partial failure does not leave the system in an inconsistent state. These are the tests that catch the regressions a happy-path suite never would.

The entire performance benchmarking suite runs with no external service; every dependency is either in-process or stubbed, so every case is reproducible in CI. That is what lets the sample collector failure-mode assertions live as tests rather than as a manual checklist that drifts out of date.

# run the performance benchmarking test suite
pnpm vitest run

Deployment and operations

Run at least two performance benchmarking instances behind a load balancer and share state in Redis. Set resource limits so performance benchmarking overload surfaces as backpressure rather than failure.

Run at least two stateless instances behind a load balancer, and move any shared state into a shared store so the performance benchmarking behavior is global rather than per instance. Size limits so the system sheds load before it runs out of memory, because a performance benchmarking system that OOMs under load fails in a way that is hard to reason about.

Drain on deploy: stop accepting new work, let in-flight performance benchmarking operations finish, then exit. Because the system holds no state that cannot be reconstructed, a rollout is a matter of cycling instances one at a time while watching the error rate.

When a dependency is unavailable, decide explicitly whether the performance benchmarking system fails closed or degrades. Leaving that choice to chance is how a dependency outage becomes a silent data loss, so make the choice visible in code and in a test.

Security considerations

Validate all performance benchmarking inputs at the boundary, reject oversized payloads before buffering them, and keep secrets in the environment rather than the repository.

The performance benchmarking system is a choke point, so it is the right place to reject malformed input before it reaches warmup runner or sample collector. Validate the shape and the size at the edge, and let the inner components assume what they receive is well-formed.

Never log secrets or full payloads; log enough to localize a failure without recording data that should not be on disk. Treat any external call as untrusted and validate its response before using it, because a performance benchmarking system that trusts its inputs blindly is one malformed response away from corruption.

Keep dependencies pinned and run an audit in CI. A transitive dependency with a known vulnerability is still a vulnerability in your performance benchmarking system, and the lockfile is what makes the audit meaningful.

Observability

Emit one structured event per performance benchmarking operation with the inputs’ shape, the outcome, and a correlation id. Track error rate and latency per component and alert on a step change rather than a single spike.

The correlation id is the thread that ties a user-visible symptom to the performance benchmarking decision and to the component log entry for the same operation. Without it, an investigation starts from a timestamp and a guess.

Watch the sample collector failure rate separately from the overall error rate. A jump in sample collector failures usually means a noisy environment where variance swamps the regression signal and the reporter flags random fluctuations, not a bug in warmup runner, and the two have different owners and different fixes.

Chart the share of performance benchmarking operations that shed, queue, or succeed. A change in the mix points at the cause faster than an aggregate latency can, because each class has a different root.

Troubleshooting

A noisy environment where variance swamps the regression signal and the reporter flags random fluctuations. Reproduce with a minimal fixture, then add a guard test so the performance benchmarking system rejects the condition loudly.

Performance regresses under performance benchmarking load. Profile with the sampling analyzer, locate the hot path in warmup runner, and add a targeted benchmark to lock the regression.

Keep the performance benchmarking dashboards separate from dependency dashboards. When an operation fails, the correlation id lets you decide in seconds whether the fault is benchmark loader, warmup runner, or sample collector, instead of guessing from a timeout.

Limitations and next steps

• Single-region; multi-region replication is out of scope.
• The benchmark loader is in-process and not horizontally shared by default.
• No auth layer; assume deployment behind an edge that enforces identity.

The performance benchmarking system is single-region by design; extending to multi-region means making benchmark loader shareable across regions, which is a real change to the consistency model rather than a config flag. The sample collector is in-process, which is correct for one instance but must move to a shared store before scaling out.

There is no authentication or authorization layer; the system assumes it is deployed behind an edge that enforces identity. Adding one is straightforward but out of scope for this build, and doing it here would distract from the performance benchmarking correctness that is the point.

The natural next step, once the performance benchmarking pipeline is trusted in production, is to add a short-TTL cache for idempotent reads and a richer stat analyzer that exports histograms rather than counters. Both are additive and neither changes the contracts this build establishes.