Aurora Serverless v2 update
- AWS described Aurora Serverless v2 improvements delivering faster performance and improved scaling toward zero compute. (x.com) - The changes promise lower latency under variable loads and more efficient cost behavior for intermittent databases. (x.com) - The social summary says customers should see smoother autoscaling for bursty workloads with fewer manual tuning steps. (x.com)
A serverless database is supposed to add computing power when traffic spikes and shrink when traffic fades. Amazon Web Services said Aurora Serverless v2 now does that up to 30% faster and can finish some workloads 32.9% sooner on its latest platform version. (aws.amazon.com) Aurora is Amazon Web Services’ managed relational database, and its “serverless” mode lets customers set a floor and ceiling for compute instead of picking a fixed machine size. Amazon measures that compute in Aurora Capacity Units, or ACUs, with 1 ACU equal to about 2 gibibytes of memory plus matching central processing and networking capacity. (docs.aws.amazon.com 1) (docs.aws.amazon.com 2) The new AWS blog post, published April 20, 2026, said Aurora Serverless v2 on platform version 3 can scale up as much as 45.0% faster under load. AWS said the update is available at no extra charge and is aimed at databases that swing between idle periods and sudden bursts of traffic. (aws.amazon.com) AWS had already pushed Aurora Serverless v2 closer to a “true serverless” model in November 2024, when it added support for scaling down to 0 ACUs and automatically pausing after inactivity. The company’s documentation says customers are not charged for instance capacity while a database is paused, though storage charges still apply. (aws.amazon.com) (docs.aws.amazon.com) That matters for teams running development databases, internal tools, or consumer apps with uneven traffic, where fixed database instances can sit half-empty for hours. AWS lists variable workloads, infrequent use, and mixed clusters with both provisioned and serverless instances among Aurora Serverless v2’s main use cases. (docs.aws.amazon.com) (aws.amazon.com) The tradeoff is that customers still have to choose a minimum and maximum capacity range for each cluster, and scaling speed depends in part on how wide that range is. AWS documentation also says the allowed minimum depends on the Aurora engine version, even though the overall v2 range now runs from 0 to 256 ACUs in 0.5-ACU steps. (docs.aws.amazon.com 1) (docs.aws.amazon.com 2) AWS is pitching the latest change as a way to cut manual tuning for bursty applications without giving up Aurora’s higher-end database features. The company’s update keeps the original sales pitch intact: pay for more compute when demand arrives, and let the database shrink back toward zero when it does not. (aws.amazon.com)
