Aurora Serverless
What you'll learn: How Aurora Serverless v2 automatically scales relational database capacity on-demand using granular capacity units.
What Is Aurora Serverless?
Aurora Serverless is AWS's auto-scaling configuration for Amazon Aurora (MySQL/PostgreSQL-compatible). Unlike traditional databases where you provision fixed compute instances, Aurora Serverless automatically scales database capacity up or down based on actual application demand. Think of it like a taxi that only charges for the distance you travel, versus owning a car that sits idle most of the time.
Aurora Serverless v2: Core Mechanics
Aurora Serverless v2 measures capacity in Aurora Capacity Units (ACUs). Each ACU represents approximately 2 GB of memory plus corresponding CPU and networking. You set a minimum and maximum ACU range (e.g., 0.5 to 16 ACUs), and Aurora scales within that range in increments as small as 0.5 ACU.
Scaling happens instantly (within a second), without connection drops. This is a massive improvement over v1, which paused briefly during scaling events. The database monitors CPU utilization, connections, and workload patterns to predict and respond to demand changes proactively.
When to Use Aurora Serverless
Aurora Serverless shines when:
- Workload variability is high (spiky traffic, dev/test environments)
- You want hands-off capacity management without over-provisioning
- You need relational ACID guarantees that NoSQL databases like DynamoDB can't provide
- You're building serverless applications where compute scales automatically but need matching database scaling
Avoid it for extremely predictable, constant high-throughput workloads where provisioned instances might be more cost-effective.
Aurora vs DynamoDB
While DynamoDB offers millisecond latency and infinite scale, Aurora Serverless provides familiar SQL, complex joins, and transactions—crucial when your data model requires relational integrity.
Key Takeaway: Aurora Serverless v2 auto-scales relational database capacity using fractional ACUs with sub-second response times, making it ideal for variable workloads requiring SQL capabilities.