Aurora Serverless v2 vs RDS: when to use which

#devops #aws #rds #aurora

Aurora Serverless v2 vs RDS: when to use which (with real cost comparisons)

The key difference

RDS (provisioned):    Fixed capacity, fixed cost, consistent performance
Aurora Serverless v2: Scales between min/max ACU, minimum 0.5 ACU always running

Cost comparison (PostgreSQL, us-east-1)

Scenario 1: Low-traffic dev database (constant)
  RDS db.t4g.micro:          $11.52/month
  Aurora Serverless v2 0.5:  $43.20/month
  Verdict: RDS wins 4:1

Scenario 2: Predictable production (sized for peak)
  Aurora Serverless v2:      ~$144/month
  RDS db.t4g.medium:         ~$46.80/month
  Verdict: RDS wins again

Scenario 3: Truly unpredictable spikes (rare 16 ACU bursts)
  Aurora Serverless v2:      ~$50-80/month average
  RDS sized for 16 ACU:      $346/month constantly
  Verdict: Aurora Serverless v2 wins

When Aurora Serverless v2 is right

Truly unpredictable workloads (traffic that 10xs without warning)
Multi-tenant SaaS (many per-tenant databases, mostly idle)
Dev databases where you want near-zero cost

When standard RDS is better

Predictable steady-state load (almost everything)
Cost-sensitive: db.t4g.micro is dramatically cheaper
You need specific PostgreSQL extensions
Latency-sensitive: RDS has more consistent p99

Terraform for Aurora Serverless v2

resource "aws_rds_cluster" "main" {
  engine = "aurora-postgresql"; engine_mode = "provisioned"
  serverlessv2_scaling_configuration { min_capacity = 0.5; max_capacity = 16 }
}
resource "aws_rds_cluster_instance" "main" {
  instance_class = "db.serverless"  # Required for Serverless v2
  cluster_identifier = aws_rds_cluster.main.id
}