Redis Caching Strategies for APIs

Caching turns read-heavy endpoints from database bottlenecks into millisecond responses—when TTLs, invalidation, and stampede protection are deliberate.

Cache-aside pattern

async getPost(slug: string): Promise<Post> {
  const cached = await this.redis.get(`post:${slug}`);
  if (cached) return JSON.parse(cached);
  const post = await this.prisma.read.post.findUnique({ where: { slug } });
  await this.redis.set(`post:${slug}`, JSON.stringify(post), 'EX', 300);
  return post;
}

TTL policy matrix

Stampede protection

Use probabilistic early expiration or request coalescing (single-flight) when hot keys expire. A viral post should not trigger 10,000 simultaneous database queries.

When not to cache

• Rapidly mutating personalized dashboards
• Financial balances requiring strong consistency
• Responses larger than 1MB—consider CDN instead

Monitor hit ratio per endpoint. Below 60% on intentionally cached routes means TTL or key design needs rework.

Redis caching is a product decision as much as an infrastructure choice—stale data tolerance must be explicit.