阅读中文版本

Build a Feature Flag Service

Feature Flag Service — a complete walkthrough of feature toggling, covering flag store, rule evaluator, rollout bucket, and testing.

feature_toggle_service/
├── src/index.ts
├── src/flag-store.ts
├── src/rule-evaluator.ts
├── tests/feature_toggle_service.test.ts
└── package.json

// Flag store: feature toggling core
export function flagstore(input: FlagstoreInput): FlagstoreResult {
  // Build the flag store that anchors feature toggling and defines the core data shape.
  return processFlagstore(input);
}

// Rule evaluator — handles feature toggling Add the rule evaluator so
export class Ruleevaluator {
  constructor(private opts: RuleevaluatorOptions) {}
  async run(ctx: Context): Promise<RuleevaluatorOutput> {
    return this.execute(ctx);
  }
}

// Rollout bucket: feature toggling stage
export const rolloutbucket = (next: Next) => async (req: Request) => {
  const out = await transform(req);
  return next(out);
};

// Snapshot cache for feature toggling
export async function snapshotcache(stream: AsyncIterable<Chunk>): Promise<Aggregate> {
  let acc = init();
  for await (const c of stream) acc = merge(acc, c);
  return acc;
}

// Change watcher: feature toggling finalization
export function changewatcher(state: State): Result {
  validate(state);
  return finalize(state);
}

Project overview

This project builds a feature toggling system whose reason for existing is serving flag evaluations with targeting rules and gradual rollouts without a flag change becoming a deploy. The temptation in feature toggling 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 rollout hash that clusters identifiers and gives one cohort far more traffic than the percentage implies as the first thing designed, not the last.

You will implement Flag store, Rule evaluator, Rollout bucket, Snapshot cache, and Change watcher, each as a module with one responsibility and a test that proves it in isolation. The feature toggling pipeline is then the wiring that composes them, and the integration test is where the composition is shown to survive a rollout hash that clusters identifiers and gives one cohort far more traffic than the percentage implies.

The walkthrough is written for a developer who has used a feature toggling 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 flag store that returns the right shape, a rule evaluator that rejects bad input, an integration test that goes green — so you can stop at any point and still have a feature toggling system that works for the cases covered so far.

What you will build

The finished feature toggling system exposes five capabilities, each with a focused test:

• Flag store
• Rule evaluator
• Rollout bucket
• Snapshot cache
• Change watcher

Flag store is the foundation that every later stage assumes; Rule evaluator is where feature toggling correctness lives; Rollout bucket handles a rollout hash that clusters identifiers and gives one cohort far more traffic than the percentage implies; Snapshot cache makes the system observable; and Change watcher is the graceful-degradation layer that keeps overload from becoming corruption.

The build keeps the feature toggling 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 Flag store, Rule evaluator, and Rollout bucket, with Snapshot cache tying them together. Data flows as a set of flags each with rules keyed by attribute, plus a rollout percentage evaluated by hashing an identifier into a bucket.

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 Rollout bucket does not need to stand up Flag store and Rule evaluator to exercise the interesting branch.

The data model is deliberately explicit. A set of flags each with rules keyed by attribute, plus a rollout percentage evaluated by hashing an identifier into a bucket 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 feature toggling 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 Rollout bucket is designed before the happy path is finished, because a rollout hash that clusters identifiers and gives one cohort far more traffic than the percentage implies is the case that distinguishes a feature toggling 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 feature toggling setup is not blocked by a missing tool:

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

The runtime dependencies are chosen to keep the feature toggling datapath legible:

• fastify (^5.0.0) — Evaluation API
• redis (^4.7.0) — Flag store
• xxhash-wasm (^0.4.0) — Stable bucketing hash

# setup for feature-toggle-service
pnpm init
pnpm add fastify redis xxhash-wasm
pnpm add -D typescript tsx vitest

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

Directory structure

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

feature_toggle_service/
├── src/index.ts
├── src/flag-store.ts
├── src/rule-evaluator.ts
├── tests/feature_toggle_service.test.ts
└── package.json

Phase 1: Flag store

Build the flag store that anchors feature toggling and defines the core data shape.

// Flag store: feature toggling core
export function flagstore(input: FlagstoreInput): FlagstoreResult {
  // Build the flag store that anchors feature toggling and defines the core data shape.
  return processFlagstore(input);
}

After this phase the flag store works in isolation and is covered by a unit test. The flag store 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 flag store as the single source of truth for the feature toggling data shape. Validate inputs at the boundary and reject early, so a malformed value never reaches Rule evaluator or Rollout bucket where its effect would be harder to localize.

Phase 2: Rule evaluator

Add the rule evaluator so feature toggling can serving flag evaluations with correctly.

// Rule evaluator — handles feature toggling Add the rule evaluator so
export class Ruleevaluator {
  constructor(private opts: RuleevaluatorOptions) {}
  async run(ctx: Context): Promise<RuleevaluatorOutput> {
    return this.execute(ctx);
  }
}

After this phase the rule evaluator works in isolation and is covered by a unit test. This is where feature toggling correctness lives: the rule evaluator is what makes serving flag evaluations with targeting rules and gradual rollouts without a flag change becoming a deploy true, and its test is the proof.

Keep the rule evaluator 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 rule evaluator about the transformation.

Phase 3: Rollout bucket

Wire the rollout bucket to handle a rollout hash that without losing data.

// Rollout bucket: feature toggling stage
export const rolloutbucket = (next: Next) => async (req: Request) => {
  const out = await transform(req);
  return next(out);
};

After this phase the rollout bucket works in isolation and is covered by a unit test. The rollout bucket is the difference between a feature toggling system that works on the happy path and one that survives a rollout hash that clusters identifiers and gives one cohort far more traffic than the percentage implies.

Design the failure before the success. Decide what rollout bucket does when a rollout hash that clusters identifiers and gives one cohort far more traffic than the percentage implies: reject, retry, shed, or degrade — and write that test first. A feature toggling system whose failure behavior is unspecified will pick one at random under pressure, and it is usually the wrong one.

Phase 4: Snapshot cache

Integrate the snapshot cache to make feature toggling observable and testable end to end.

// Snapshot cache for feature toggling
export async function snapshotcache(stream: AsyncIterable<Chunk>): Promise<Aggregate> {
  let acc = init();
  for await (const c of stream) acc = merge(acc, c);
  return acc;
}

After this phase the snapshot cache works in isolation and is covered by a unit test. Observability is not a finishing touch; it is how you know the feature toggling 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 snapshot cache is what turns “something is slow” into “the rule evaluator is slow for inputs matching this shape,” which is the difference between an alert and an investigation.

Phase 5: Change watcher

Finalize the change watcher so the feature toggling system degrades gracefully under overload.

// Change watcher: feature toggling finalization
export function changewatcher(state: State): Result {
  validate(state);
  return finalize(state);
}

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

Testing

Each phase has a focused test: flag store, rule evaluator, and rollout bucket are exercised in isolation, then an integration test drives the full feature toggling pipeline and asserts the a rollout hash that case is handled.

Start with unit tests that pin each component’s contract: the flag store returns the documented shape, the rule evaluator transforms correctly, and the rollout bucket 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 feature toggling scenario end to end. Assert the happy path first, then deliberately trigger a rollout hash that clusters identifiers and gives one cohort far more traffic than the percentage implies and assert the system degrades rather than corrupts. Because the components are pure where practical, the integration test stays deterministic rather than flaky.

A feature toggling 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 feature toggling 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 rollout bucket failure-mode assertions live as tests rather than as a manual checklist that drifts out of date.

# run the feature toggling test suite
pnpm vitest run

Deployment and operations

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

The feature toggling system is a choke point, so it is the right place to reject malformed input before it reaches rule evaluator or rollout bucket. 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 feature toggling 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 feature toggling system, and the lockfile is what makes the audit meaningful.

Observability

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

Watch the rollout bucket failure rate separately from the overall error rate. A jump in rollout bucket failures usually means a rollout hash that clusters identifiers and gives one cohort far more traffic than the percentage implies, not a bug in rule evaluator, and the two have different owners and different fixes.

Chart the share of feature toggling 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 rollout hash that clusters identifiers and gives one cohort far more traffic than the percentage implies. Reproduce with a minimal fixture, then add a guard test so the feature toggling system rejects the condition loudly.

Performance regresses under feature toggling load. Profile with the sampling analyzer, locate the hot path in rule evaluator, and add a targeted benchmark to lock the regression.

Keep the feature toggling dashboards separate from dependency dashboards. When an operation fails, the correlation id lets you decide in seconds whether the fault is flag store, rule evaluator, or rollout bucket, instead of guessing from a timeout.

Limitations and next steps

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

The feature toggling system is single-region by design; extending to multi-region means making flag store shareable across regions, which is a real change to the consistency model rather than a config flag. The rollout bucket 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 feature toggling correctness that is the point.

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