Amazon Aurora Serverless

What it is

Amazon Aurora Serverless is an on-demand, automatically scaling capacity option for Amazon Aurora. Instead of choosing a DB instance class, you set a minimum and maximum capacity, and Aurora adjusts each writer and reader within that range as the load changes. You pay for the capacity used, per second.

It is still Aurora: the same MySQL- and PostgreSQL-compatible engines, the same shared cluster volume across three AZ (Availability Zone)An isolated location within an AWS Region, made up of one or more data centers with their own power, cooling, and networking. The AZs in a Region are linked by fast, low-latency connections, so spreading resources across them protects against the failure of one site.Glossary entry, and the same replicas, FailoverSwitching automatically to a standby resource when the primary one fails, such as a standby database in Amazon RDS or a backup site chosen by Route 53 failover routing.Glossary entry, backups, and Global Database.

Choose it for variable, unpredictable, or intermittent workloads, such as new applications, development and test databases, multi-tenant SaaS databases, and applications with busy and quiet periods. For a steady, predictable load, provisioned Aurora instances are often cheaper.

Which version?

The exam may say Aurora Serverless v1 or v2. Aurora Serverless v1 reached end of life on March 31, 2025, and remaining v1 clusters were scheduled for automatic upgrade. In April 2026, Aurora Serverless v2 was renamed Aurora serverless, so current AWS documentation uses that name. This tutorial describes the current version. Questions written for v1 may describe a cluster that pauses when idle; the current version does this when its minimum capacity is 0 ACUs.

Concepts

Aurora capacity unit (ACU): about 2 GiB of memory with matching CPU and networking.
Capacity range: from 0 to 256 ACUs, set as a minimum and maximum for the cluster. Capacity changes in steps as small as half an ACU, without dropping connections.
Scale to zero: with a minimum of 0 ACUs, an instance with no user connections for a time you set pauses automatically. You pay nothing for instance capacity while it is paused (storage is still charged), and resuming takes about 15 seconds, so set client timeouts longer than that.
Writers and readers: a cluster can have serverless writers and readers, up to 15 readers, in several Availability Zones.
Promotion tiers: readers in tiers 0 and 1 scale with the writer, so they are ready to take over on failover. Readers in tiers 2 to 15 scale independently for read-heavy spikes.
Mixed-configuration cluster: combine provisioned and serverless instances in one cluster, for example a large provisioned writer with serverless readers.
RDS Data API (Application programming interface)A defined way for one program to ask another for data or actions, such as a set of HTTPS requests and responses. Every AWS service is controlled through its API, which the console, the CLI, and the SDKs call for you.Glossary entry: run SQL (Structured Query Language)The standard language for querying and changing data in relational databases. Analytics services such as Amazon Athena and Amazon Redshift also use SQL.Glossary entry over a secure HTTPS (HTTP Secure)HTTP sent over a TLS-encrypted connection, so that requests and responses cannot be read or changed on the way, and the server proves its identity with a certificate.Glossary entry endpoint through the AWS SDK (Software development kit)A set of libraries for a programming language that wraps a service's API and handles details such as signing requests and retrying. AWS SDKs exist for Python (Boto3), JavaScript, Java, Go, and other languages.Glossary entry, with credentials in Secrets Manager and no persistent connection. It suits Lambda and other short-lived clients.

Security, resilience, performance, and cost

Security

The same controls as Aurora: SubnetA range of IP addresses in a VPC, inside one Availability Zone. A subnet is public when its route table sends internet traffic to an internet gateway, and private when it does not.Glossary entry and Security groupA stateful virtual firewall for network interfaces, such as an EC2 instance's, whose rules can only allow traffic. Return traffic for allowed connections is allowed automatically, and a rule can refer to another security group instead of IP addresses.Glossary entry, KMS Encryption at restEncrypting data where it is stored, such as on disks, in databases, or in backups, so that it cannot be read without the key. On AWS it usually uses keys in AWS KMS.Glossary entry, TLS (Transport Layer Security)The protocol that encrypts traffic between a client and a server and lets the server prove its identity with a certificate. It replaced SSL, though the older name is still common; HTTPS is HTTP over TLS.Glossary entry in transit, IAM database authentication, and Secrets Manager for credentials (also used by the Data API).

Resilience

Put a reader in another Availability Zone, in promotion tier 0 or 1, so a failover target is always sized to match the writer.
Continuous backups with PITR (Point-in-time recovery)Restoring a database to its state at a chosen moment within a retention period, using continuous backups. Amazon RDS, Amazon Aurora, Amazon DynamoDB, and AWS Backup support it.Glossary entry, cloning, and Global Database work as they do for provisioned Aurora.

Performance

Set the minimum high enough for the working set to stay in memory and for fast responses after quiet periods; set the maximum to cap cost and protect against runaway load.
A paused instance adds about 15 seconds to the first connection, so avoid a minimum of 0 for LatencyThe time between sending a request and receiving the response, usually measured in milliseconds. Distance and processing add to it; caching and serving users from nearby locations reduce it.Glossary entry-sensitive production traffic.
For short-lived connections, use RDS Proxy or the Data API.

Cost

Pay per ACU-hour, billed per second, plus storage, I/O (with Aurora Standard), and backup storage.
A minimum of 0 ACUs suits development and test databases that sit idle for long periods.
Steady usage can be covered by Database Savings Plans; for a constant load, compare with provisioned instances and Reserved DB Instances.

Monitoring

The ServerlessDatabaseCapacity and ACUUtilization CloudWatch metrics show how much capacity is used and whether the range is right. Database Insights and CloudWatch Logs work as for provisioned Aurora.

Serverless or provisioned Aurora?

If the requirement is	Choose
Unpredictable or spiky load, no capacity planning	Aurora Serverless
Development or test database that is idle most of the time	Aurora Serverless with a minimum of 0 ACUs
Many tenant databases with different, changing loads	Aurora Serverless
Steady, predictable load at the lowest cost	Provisioned Aurora with Reserved DB Instances
A larger writer than the serverless maximum, with flexible readers	Mixed-configuration cluster
A NoSQLDatabases that store data in models other than relational tables, such as key-value, document, wide-column, or graph, usually to scale out easily or to handle data whose shape varies. Amazon DynamoDB is a NoSQL database.Glossary entry, serverless key-value database	Amazon DynamoDB

Reverse mindmap: when the question points here

Is it Aurora Serverless?

Yes, when the question says
- Relational database with unpredictable or intermittent load leads to Aurora Serverless
- No instance sizing, capacity adjusts automatically leads to Aurora Serverless
- Pay nothing for compute while the database is idle leads to Scale to 0 ACUs
Know how it works
- Unit of capacity leads to ACU, about 2 GiB of memory
- Readers ready for failover leads to Promotion tier 0 or 1
- Readers that scale on their own leads to Promotion tier 2 to 15
- SQL over HTTPS from Lambda leads to RDS Data API
No, when the question says
- Steady load at the lowest cost leads to Provisioned Aurora with reservations
- Oracle or SQL Server leads to Amazon RDS
- Key-value, serverless at any scale leads to DynamoDB

Key takeaways

Aurora Serverless is a capacity option for Aurora: set a range of ACUs and Aurora scales each instance within it.
One ACU is about 2 GiB of memory; the range is 0 to 256 ACUs, in half-ACU steps.
With a minimum of 0, idle instances pause and cost nothing for compute; resuming takes about 15 seconds.
Readers in promotion tiers 0 and 1 track the writer's size; tiers 2 to 15 scale independently.
v1 reached end of life on March 31, 2025; v2 was renamed "Aurora serverless" in April 2026.
For steady load, provisioned Aurora is often cheaper.