阅读中文版本

Build a Sampling Profiler Analyzer

Sampling Profiler Analyzer — a complete walkthrough of profile analysis, covering sample ingest, stack merger, flame builder, and testing.

pprof_sampling_analyzer/
├── src/index.ts
├── src/sample-ingest.ts
├── src/stack-merger.ts
├── tests/pprof_sampling_analyzer.test.ts
└── package.json

// Sample ingest: profile analysis core
export function sampleingest(input: SampleingestInput): SampleingestResult {
  // Build the sample ingest that anchors profile analysis and defines the core data shape.
  return processSampleingest(input);
}

// Stack merger — handles profile analysis Add the stack merger so
export class Stackmerger {
  constructor(private opts: StackmergerOptions) {}
  async run(ctx: Context): Promise<StackmergerOutput> {
    return this.execute(ctx);
  }
}

// Flame builder: profile analysis stage
export const flamebuilder = (next: Next) => async (req: Request) => {
  const out = await transform(req);
  return next(out);
};

// Symbolizer for profile analysis
export async function symbolizer(stream: AsyncIterable<Chunk>): Promise<Aggregate> {
  let acc = init();
  for await (const c of stream) acc = merge(acc, c);
  return acc;
}

// Diff reporter: profile analysis finalization
export function diffreporter(state: State): Result {
  validate(state);
  return finalize(state);
}

Project overview

This project builds a profile analysis system whose reason for existing is aggregating stack samples into a flame graph that reveals where a program spends its time. The temptation in profile analysis 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 sampling rate so low that hot paths are lost in noise and the flame graph misleads optimization effort as the first thing designed, not the last.

You will implement Sample ingest, Stack merger, Flame builder, Symbolizer, and Diff reporter, each as a module with one responsibility and a test that proves it in isolation. The profile analysis pipeline is then the wiring that composes them, and the integration test is where the composition is shown to survive a sampling rate so low that hot paths are lost in noise and the flame graph misleads optimization effort.

The walkthrough is written for a developer who has used a profile analysis 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 sample ingest that returns the right shape, a stack merger that rejects bad input, an integration test that goes green — so you can stop at any point and still have a profile analysis system that works for the cases covered so far.

What you will build

The finished profile analysis system exposes five capabilities, each with a focused test:

• Sample ingest
• Stack merger
• Flame builder
• Symbolizer
• Diff reporter

Sample ingest is the foundation that every later stage assumes; Stack merger is where profile analysis correctness lives; Flame builder handles a sampling rate so low that hot paths are lost in noise and the flame graph misleads optimization effort; Symbolizer makes the system observable; and Diff reporter is the graceful-degradation layer that keeps overload from becoming corruption.

The build keeps the profile analysis 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 Sample ingest, Stack merger, and Flame builder, with Symbolizer tying them together. Data flows as a frequency table of stack traces, merged into a tree where node weight is the share of samples in that subtree.

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 Flame builder does not need to stand up Sample ingest and Stack merger to exercise the interesting branch.

The data model is deliberately explicit. A frequency table of stack traces, merged into a tree where node weight is the share of samples in that subtree 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 profile analysis 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 Flame builder is designed before the happy path is finished, because a sampling rate so low that hot paths are lost in noise and the flame graph misleads optimization effort is the case that distinguishes a profile analysis 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 profile analysis setup is not blocked by a missing tool:

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

The runtime dependencies are chosen to keep the profile analysis datapath legible:

• fastify (^5.0.0) — Profile API
• pprof (^3.0.0) — Sample format

# setup for pprof-sampling-analyzer
pnpm init
pnpm add fastify pprof
pnpm add -D typescript tsx vitest

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

Directory structure

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

pprof_sampling_analyzer/
├── src/index.ts
├── src/sample-ingest.ts
├── src/stack-merger.ts
├── tests/pprof_sampling_analyzer.test.ts
└── package.json

Phase 1: Sample ingest

Build the sample ingest that anchors profile analysis and defines the core data shape.

// Sample ingest: profile analysis core
export function sampleingest(input: SampleingestInput): SampleingestResult {
  // Build the sample ingest that anchors profile analysis and defines the core data shape.
  return processSampleingest(input);
}

After this phase the sample ingest works in isolation and is covered by a unit test. The sample ingest 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 sample ingest as the single source of truth for the profile analysis data shape. Validate inputs at the boundary and reject early, so a malformed value never reaches Stack merger or Flame builder where its effect would be harder to localize.

Phase 2: Stack merger

Add the stack merger so profile analysis can aggregating stack samples into correctly.

// Stack merger — handles profile analysis Add the stack merger so
export class Stackmerger {
  constructor(private opts: StackmergerOptions) {}
  async run(ctx: Context): Promise<StackmergerOutput> {
    return this.execute(ctx);
  }
}

After this phase the stack merger works in isolation and is covered by a unit test. This is where profile analysis correctness lives: the stack merger is what makes aggregating stack samples into a flame graph that reveals where a program spends its time true, and its test is the proof.

Keep the stack merger 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 stack merger about the transformation.

Phase 3: Flame builder

Wire the flame builder to handle a sampling rate so without losing data.

// Flame builder: profile analysis stage
export const flamebuilder = (next: Next) => async (req: Request) => {
  const out = await transform(req);
  return next(out);
};

After this phase the flame builder works in isolation and is covered by a unit test. The flame builder is the difference between a profile analysis system that works on the happy path and one that survives a sampling rate so low that hot paths are lost in noise and the flame graph misleads optimization effort.

Design the failure before the success. Decide what flame builder does when a sampling rate so low that hot paths are lost in noise and the flame graph misleads optimization effort: reject, retry, shed, or degrade — and write that test first. A profile analysis system whose failure behavior is unspecified will pick one at random under pressure, and it is usually the wrong one.

Phase 4: Symbolizer

Integrate the symbolizer to make profile analysis observable and testable end to end.

// Symbolizer for profile analysis
export async function symbolizer(stream: AsyncIterable<Chunk>): Promise<Aggregate> {
  let acc = init();
  for await (const c of stream) acc = merge(acc, c);
  return acc;
}

After this phase the symbolizer works in isolation and is covered by a unit test. Observability is not a finishing touch; it is how you know the profile analysis 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 symbolizer is what turns “something is slow” into “the stack merger is slow for inputs matching this shape,” which is the difference between an alert and an investigation.

Phase 5: Diff reporter

Finalize the diff reporter so the profile analysis system degrades gracefully under overload.

// Diff reporter: profile analysis finalization
export function diffreporter(state: State): Result {
  validate(state);
  return finalize(state);
}

After this phase the diff reporter works in isolation and is covered by a unit test. The diff reporter is what keeps overload from becoming corruption: when the profile analysis 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 diff reporter — shed, queue, or backpressure — and make it explicit in code and in a test. Leaving the behavior to chance is how a profile analysis traffic spike becomes a data loss incident.

Testing

Each phase has a focused test: sample ingest, stack merger, and flame builder are exercised in isolation, then an integration test drives the full profile analysis pipeline and asserts the a sampling rate so case is handled.

Start with unit tests that pin each component’s contract: the sample ingest returns the documented shape, the stack merger transforms correctly, and the flame builder 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 profile analysis scenario end to end. Assert the happy path first, then deliberately trigger a sampling rate so low that hot paths are lost in noise and the flame graph misleads optimization effort and assert the system degrades rather than corrupts. Because the components are pure where practical, the integration test stays deterministic rather than flaky.

A profile analysis 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 profile analysis 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 flame builder failure-mode assertions live as tests rather than as a manual checklist that drifts out of date.

# run the profile analysis test suite
pnpm vitest run

Deployment and operations

Run at least two profile analysis instances behind a load balancer and share state in Redis. Set resource limits so profile analysis 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 profile analysis behavior is global rather than per instance. Size limits so the system sheds load before it runs out of memory, because a profile analysis 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 profile analysis 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 profile analysis 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 profile analysis inputs at the boundary, reject oversized payloads before buffering them, and keep secrets in the environment rather than the repository.

The profile analysis system is a choke point, so it is the right place to reject malformed input before it reaches stack merger or flame builder. 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 profile analysis 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 profile analysis system, and the lockfile is what makes the audit meaningful.

Observability

Emit one structured event per profile analysis 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 profile analysis decision and to the component log entry for the same operation. Without it, an investigation starts from a timestamp and a guess.

Watch the flame builder failure rate separately from the overall error rate. A jump in flame builder failures usually means a sampling rate so low that hot paths are lost in noise and the flame graph misleads optimization effort, not a bug in stack merger, and the two have different owners and different fixes.

Chart the share of profile analysis 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 sampling rate so low that hot paths are lost in noise and the flame graph misleads optimization effort. Reproduce with a minimal fixture, then add a guard test so the profile analysis system rejects the condition loudly.

Performance regresses under profile analysis load. Profile with the sampling analyzer, locate the hot path in stack merger, and add a targeted benchmark to lock the regression.

Keep the profile analysis dashboards separate from dependency dashboards. When an operation fails, the correlation id lets you decide in seconds whether the fault is sample ingest, stack merger, or flame builder, instead of guessing from a timeout.

Limitations and next steps

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

The profile analysis system is single-region by design; extending to multi-region means making sample ingest shareable across regions, which is a real change to the consistency model rather than a config flag. The flame builder 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 profile analysis correctness that is the point.

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