The Cache-Aside Pattern

What you'll walk away with

Explain the core ideas behind The Cache-Aside Pattern
Run the sample Redis command or code and verify its output
Apply the technique correctly to the Tutorial Platform and production scenarios

Build the mental model

With cache-aside, the application reads Redis first, falls back to PostgreSQL on a miss, and stores the result with a TTL. On writes, commit the database change before deleting related cache keys. High hit rates help latency, but stale data, stampedes, and cold starts must be designed explicitly.

Connect it to a real scenario

Cache tutorial details under `tutorial:42:v1`. On a miss, use parameterized SQL, serialize the result, and set it with a jittered TTL. Delete the key after a successful update commit. If Redis is unavailable, fall back to the database with overload protection.

Read from cache, then DB, then populate cache
Commit DB writes before deleting cache
Use TTL jitter and a deliberate fallback policy

Try the working example

typescript

async function getTutorial(id: number) {
  const key = `tutorial:${id}:v1`;
  const cached = await redis.get(key);
  if (cached) return JSON.parse(cached);

  const tutorial = await db.oneOrNone(
    'SELECT id, title, summary FROM tutorials WHERE id = $1', [id]
  );
  if (tutorial) await redis.set(key, JSON.stringify(tutorial), { EX: 300 + Math.floor(Math.random() * 30) });
  return tutorial;
}

You should see

The first request reads PostgreSQL; later requests hit Redis until expiry or invalidation.

5-minute try-it

Design the key scheme, TTL, and invalidation events for paginated tutorial lists.

One important caution

Deleting cache before the database update can let a concurrent reader repopulate the old value.

Redis — Client-Side Caching — Redis

Easy traps

Deleting cache before the database update can let a concurrent reader repopulate the old value.
Validate sample commands on a local or test instance with recoverable data before applying them to production Redis.

Exercise

Design the key scheme, TTL, and invalidation events for paginated tutorial lists.

You'll know it worked when: The first request reads PostgreSQL; later requests hit Redis until expiry or invalidation.