Amazon Aurora Serverless has emerged as a dynamic solution for database management, offering an on-demand, auto-scaling configuration that adapts to the fluctuating demands of workloads. This innovative service can scale up to meet the most intensive requirements and gracefully down to zero during periods of inactivity, ensuring that users only pay for what they actually consume. Particularly beneficial for agentic AI applications, which often experience unpredictable activity patterns, Aurora Serverless manages capacity seamlessly, allowing developers to focus on their applications rather than database management. Recent enhancements have led to performance improvements of up to 30%, alongside a scaling mechanism that is more attuned to workload demands, all without incurring additional costs.
Better performance out of the box
The latest iteration of Aurora Serverless, now operating on platform version 4, boasts a remarkable 30% increase in performance compared to its predecessor, platform version 3. This evolution is part of a continuous effort to enhance efficiency while keeping costs manageable. The introduction of platform versions allows users to monitor performance metrics and scaling baselines effectively. New clusters, database restores, and clones are automatically provisioned with the latest platform version available in the chosen AWS Region. Existing clusters can easily transition to the latest version through various methods, including pending maintenance actions, stopping and restarting the cluster, or utilizing blue/green deployments.
Benchmark results from a HammerDB TPROC-C test illustrate the incremental performance gains across three serverless platform versions, with a focus on New Orders per Minute (NOPM) at a user count of 1,024. The data reveals that both Aurora MySQL and Aurora PostgreSQL on platform version 4 achieve a significant increase in NOPM—between 27% and 34%—compared to platform version 3.
Faster and smarter autoscaling
Aurora Serverless employs an intelligent autoscaling mechanism that autonomously allocates CPU and memory resources in response to workload demands, while efficiently releasing them once tasks are completed. Recent updates have doubled the default scaling rate across all serverless clusters, requiring no additional configuration. A comparative analysis of scaling times from 0.5 ACU to 256 ACU indicates that the new scaling behavior achieves maximum capacity in just 22 minutes, a 45% improvement over the previous 40-minute duration.
This expedited scaling not only enhances database performance but also reduces workload completion times, yielding better results without increasing costs. For instance, a Sysbench workload test demonstrated that a platform version 4 cluster completed its tasks in 27 minutes using 109 ACUs, while platform version 3 took 37 minutes with 151 ACUs, and platform version 2 required 46 minutes with 187 ACUs. The latest version thus offers a 27% faster completion time at a 28% lower cost compared to version 3, and a 41% faster completion time at a 42% lower cost than version 2.
Moreover, platform version 4 enhances the scaling algorithm by incorporating additional metrics to inform scaling decisions. This improvement allows Aurora Serverless to respond more adeptly to resource competition among concurrent tasks, such as web applications and reporting services. New clusters will automatically benefit from these advancements, while existing clusters can upgrade to platform version 4 to leverage the improved scaling behavior.
Further testing has confirmed the efficacy of the new scaling algorithm, with platform version 4 completing a Sysbench write workload in 49 minutes while consuming only 63 ACUs. In contrast, platform version 3 required 73 minutes and 66.4 ACUs, and platform version 2 took 85 minutes with 83.6 ACUs. The enhancements in platform version 4 result in a 32.9% faster completion time with 5.1% lower ACU consumption compared to version 3, and a 42.4% faster completion time with 24.6% lower ACU consumption compared to version 2.
Find your platform version
Using the Amazon RDS console
To check your platform version, follow these steps:
- Navigate to the Amazon RDS console.
- Select your Aurora Serverless cluster.
- In the Configuration tab, locate Instance configuration.
- The Platform version field will display the current serverless platform version.
Using AWS CLI
Alternatively, you can retrieve the platform version using the AWS CLI with the following command:
aws rds-beta describe-db-clusters
--db-cluster-identifier aurora-postgresql-serverless-cluster
--query 'DBClusters[0].{
ServerlessV2PlatformVersion:ServerlessV2PlatformVersion,
DBCluster:DBClusterIdentifier,
DBEngine:Engine,
DBEngineVersion:EngineVersion
}'
--output tableThis command will provide a detailed overview of your Aurora Serverless cluster, including the current platform version.
About the authors
Special thanks
Special thanks to Jason Pedreza, Lei Zeng, and Rinisha Marar for their contributions to this work.
