Aurora serverless now scales up 12 Aurora Capacity Units (ACUs) within a second and continues scaling to 256 ACUs, an improvement aimed at handling bursty workloads like agentic AI query spikes, unpredictable SaaS batch jobs, and viral traffic. The feature is enabled by default on platform versions 3 and 4 with no configuration required. Benchmark data using Sysbench shows an Aurora serverless cluster scaling from 0.5 to 12 ACUs in under a second, and reaching throughput comparable to a provisioned db.r8g.xlarge instance within about a minute during a write-heavy workload, without disrupting in-flight queries.
Table of contents
Scale in a secondFast, non-disruptive scalingSummarySpecial thanksAbout the authorsQuestions this post answers
How fast can Aurora serverless scale up capacity now?
Aurora serverless can add 12 Aurora Capacity Units (ACUs) to its current capacity within a single second, and continues scaling up to a maximum of 256 ACUs as workload demand grows. Each ACU provides roughly 2 GiB of memory with corresponding CPU and networking. This faster scaling is enabled by default on Aurora serverless clusters running platform versions 3 and 4, requiring no configuration changes. Teams sizing serverless databases for bursty AI workloads can track infrastructure changes like this on daily.dev.
How does Aurora serverless performance compare to a provisioned instance during a traffic spike?
During a write-heavy Sysbench benchmark, an Aurora serverless cluster (4 to 16 ACUs) showed a throughput gap versus a provisioned db.r8g.xlarge instance only in the first 10 seconds, because the provisioned instance already had its 16 GiB dataset cached in shared buffer. Aurora serverless closed the gap after about one minute and then matched provisioned-level throughput for the rest of the workload. Developers weighing serverless versus provisioned databases can follow benchmark comparisons like this on daily.dev.
Do I need to enable anything to get faster Aurora serverless scaling?
No configuration changes are required; the faster scaling behavior, adding 12 ACUs per second up to 256 ACUs, is turned on by default for Aurora serverless clusters running on platform versions 3 and 4. The platform version can be confirmed in the AWS Management Console instance configuration section or via the ServerlessV2PlatformVersion parameter in the Amazon RDS API. Keeping up with default AWS platform changes like this is easier when following database updates on daily.dev.