阅读中文版本

Build an LRU Cache with TTL

LRU Cache with TTL — a complete walkthrough of bounded caching, covering hash map, recency list, ttl wheel, and testing.

lru_cache_ttl/
├── src/index.ts
├── src/hash-map.ts
├── src/recency-list.ts
├── tests/lru_cache_ttl.test.ts
└── package.json

// Hash map: bounded caching core
export function hashmap(input: HashmapInput): HashmapResult {
  // Build the hash map that anchors bounded caching and defines the core data shape.
  return processHashmap(input);
}

// Recency list — handles bounded caching Add the recency list so
export class Recencylist {
  constructor(private opts: RecencylistOptions) {}
  async run(ctx: Context): Promise<RecencylistOutput> {
    return this.execute(ctx);
  }
}

// TTL wheel: bounded caching stage
export const ttlwheel = (next: Next) => async (req: Request) => {
  const out = await transform(req);
  return next(out);
};

// Eviction policy for bounded caching
export async function evictionpolicy(stream: AsyncIterable<Chunk>): Promise<Aggregate> {
  let acc = init();
  for await (const c of stream) acc = merge(acc, c);
  return acc;
}

// Stats counter: bounded caching finalization
export function statscounter(state: State): Result {
  validate(state);
  return finalize(state);
}

Project overview

This project builds a bounded caching system whose reason for existing is evicting both the least recently used and the expired entries so the cache stays bounded and fresh. The temptation in bounded caching 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 TTL wheel with too few buckets that expires entries early or late depending on clock alignment as the first thing designed, not the last.

You will implement Hash map, Recency list, TTL wheel, Eviction policy, and Stats counter, each as a module with one responsibility and a test that proves it in isolation. The bounded caching pipeline is then the wiring that composes them, and the integration test is where the composition is shown to survive a TTL wheel with too few buckets that expires entries early or late depending on clock alignment.

The walkthrough is written for a developer who has used a bounded caching 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 hash map that returns the right shape, a recency list that rejects bad input, an integration test that goes green — so you can stop at any point and still have a bounded caching system that works for the cases covered so far.

What you will build

The finished bounded caching system exposes five capabilities, each with a focused test:

• Hash map
• Recency list
• TTL wheel
• Eviction policy
• Stats counter

Hash map is the foundation that every later stage assumes; Recency list is where bounded caching correctness lives; TTL wheel handles a TTL wheel with too few buckets that expires entries early or late depending on clock alignment; Eviction policy makes the system observable; and Stats counter is the graceful-degradation layer that keeps overload from becoming corruption.

The build keeps the bounded caching 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 Hash map, Recency list, and TTL wheel, with Eviction policy tying them together. Data flows as a hash map of entries plus a doubly linked list for recency order and a timing wheel for per-entry expiry.

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 TTL wheel does not need to stand up Hash map and Recency list to exercise the interesting branch.

The data model is deliberately explicit. A hash map of entries plus a doubly linked list for recency order and a timing wheel for per-entry expiry 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 bounded caching 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 TTL wheel is designed before the happy path is finished, because a TTL wheel with too few buckets that expires entries early or late depending on clock alignment is the case that distinguishes a bounded caching 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 bounded caching setup is not blocked by a missing tool:

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

The runtime dependencies are chosen to keep the bounded caching datapath legible:

• lru-cache (^11.0.0) — LRU primitive
• fastify (^5.0.0) — Cache API

# setup for lru-cache-ttl
pnpm init
pnpm add lru-cache fastify
pnpm add -D typescript tsx vitest

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

Directory structure

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

lru_cache_ttl/
├── src/index.ts
├── src/hash-map.ts
├── src/recency-list.ts
├── tests/lru_cache_ttl.test.ts
└── package.json

Phase 1: Hash map

Build the hash map that anchors bounded caching and defines the core data shape.

// Hash map: bounded caching core
export function hashmap(input: HashmapInput): HashmapResult {
  // Build the hash map that anchors bounded caching and defines the core data shape.
  return processHashmap(input);
}

After this phase the hash map works in isolation and is covered by a unit test. The hash map 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 hash map as the single source of truth for the bounded caching data shape. Validate inputs at the boundary and reject early, so a malformed value never reaches Recency list or TTL wheel where its effect would be harder to localize.

Phase 2: Recency list

Add the recency list so bounded caching can evicting both the least correctly.

// Recency list — handles bounded caching Add the recency list so
export class Recencylist {
  constructor(private opts: RecencylistOptions) {}
  async run(ctx: Context): Promise<RecencylistOutput> {
    return this.execute(ctx);
  }
}

After this phase the recency list works in isolation and is covered by a unit test. This is where bounded caching correctness lives: the recency list is what makes evicting both the least recently used and the expired entries so the cache stays bounded and fresh true, and its test is the proof.

Keep the recency list 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 recency list about the transformation.

Phase 3: TTL wheel

Wire the ttl wheel to handle a TTL wheel with without losing data.

// TTL wheel: bounded caching stage
export const ttlwheel = (next: Next) => async (req: Request) => {
  const out = await transform(req);
  return next(out);
};

After this phase the ttl wheel works in isolation and is covered by a unit test. The ttl wheel is the difference between a bounded caching system that works on the happy path and one that survives a TTL wheel with too few buckets that expires entries early or late depending on clock alignment.

Design the failure before the success. Decide what ttl wheel does when a TTL wheel with too few buckets that expires entries early or late depending on clock alignment: reject, retry, shed, or degrade — and write that test first. A bounded caching system whose failure behavior is unspecified will pick one at random under pressure, and it is usually the wrong one.

Phase 4: Eviction policy

Integrate the eviction policy to make bounded caching observable and testable end to end.

// Eviction policy for bounded caching
export async function evictionpolicy(stream: AsyncIterable<Chunk>): Promise<Aggregate> {
  let acc = init();
  for await (const c of stream) acc = merge(acc, c);
  return acc;
}

After this phase the eviction policy works in isolation and is covered by a unit test. Observability is not a finishing touch; it is how you know the bounded caching 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 eviction policy is what turns “something is slow” into “the recency list is slow for inputs matching this shape,” which is the difference between an alert and an investigation.

Phase 5: Stats counter

Finalize the stats counter so the bounded caching system degrades gracefully under overload.

// Stats counter: bounded caching finalization
export function statscounter(state: State): Result {
  validate(state);
  return finalize(state);
}

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

Testing

Each phase has a focused test: hash map, recency list, and ttl wheel are exercised in isolation, then an integration test drives the full bounded caching pipeline and asserts the a TTL wheel with case is handled.

Start with unit tests that pin each component’s contract: the hash map returns the documented shape, the recency list transforms correctly, and the ttl wheel 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 bounded caching scenario end to end. Assert the happy path first, then deliberately trigger a TTL wheel with too few buckets that expires entries early or late depending on clock alignment and assert the system degrades rather than corrupts. Because the components are pure where practical, the integration test stays deterministic rather than flaky.

A bounded caching 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 bounded caching 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 ttl wheel failure-mode assertions live as tests rather than as a manual checklist that drifts out of date.

# run the bounded caching test suite
pnpm vitest run

Deployment and operations

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

The bounded caching system is a choke point, so it is the right place to reject malformed input before it reaches recency list or ttl wheel. 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 bounded caching 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 bounded caching system, and the lockfile is what makes the audit meaningful.

Observability

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

Watch the ttl wheel failure rate separately from the overall error rate. A jump in ttl wheel failures usually means a TTL wheel with too few buckets that expires entries early or late depending on clock alignment, not a bug in recency list, and the two have different owners and different fixes.

Chart the share of bounded caching 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 TTL wheel with too few buckets that expires entries early or late depending on clock alignment. Reproduce with a minimal fixture, then add a guard test so the bounded caching system rejects the condition loudly.

Performance regresses under bounded caching load. Profile with the sampling analyzer, locate the hot path in recency list, and add a targeted benchmark to lock the regression.

Keep the bounded caching dashboards separate from dependency dashboards. When an operation fails, the correlation id lets you decide in seconds whether the fault is hash map, recency list, or ttl wheel, instead of guessing from a timeout.

Limitations and next steps

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

The bounded caching system is single-region by design; extending to multi-region means making hash map shareable across regions, which is a real change to the consistency model rather than a config flag. The ttl wheel 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 bounded caching correctness that is the point.

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