Database Scaling Strategies: From Single Node to Sharding

The Scaling Journey

Every successful application eventually faces the same challenge: the database becomes a bottleneck. Here’s a systematic approach to scaling your database.

Level 1: Vertical Scaling

The simplest approach — get a bigger machine.

Before: 4 CPU, 16GB RAM, 100GB SSD
After:  32 CPU, 128GB RAM, 2TB NVMe

Pros: No code changes, no complexity Cons: Expensive, has hard limits

Level 2: Read Replicas

Separate read and write traffic:

class DatabaseRouter {
  private primary: Connection;
  private replicas: Connection[];

  query(sql: string, isWrite: boolean) {
    if (isWrite) {
      return this.primary.execute(sql);
    }
    // Round-robin across replicas
    const replica = this.replicas[
      Math.floor(Math.random() * this.replicas.length)
    ];
    return replica.execute(sql);
  }
}

Level 3: Sharding

Distribute data across multiple databases:

function getShardKey(userId: string): number {
  const hash = crypto
    .createHash("md5")
    .update(userId)
    .digest("hex");
  return parseInt(hash.slice(0, 8), 16) % NUM_SHARDS;
}

Choosing the Right Strategy

Traffic	Strategy	Complexity
< 10K QPS	Vertical scaling	Low
10-100K QPS	Read replicas	Medium
100K+ QPS	Sharding	High

Conclusion

Start simple. Optimize queries first. Add indexes. Use caching. Only shard when you’ve exhausted simpler options. Premature sharding is one of the most expensive mistakes in system design.