Aurora Serverless v2 scales and bills in Aurora Capacity Units (ACUs), each about 2 GiB of memory, at roughly $0.12 per ACU-hour on Aurora Standard. You always pay for at least the minimum ACU, so the most common waste is setting it too high. Set the minimum to the real floor the workload needs, or to 0 ACUs for development and intermittent clusters to enable automatic pause after 5 minutes to 24 hours of inactivity. Set the maximum as a cost ceiling that still absorbs real peaks. Storage bills even while paused. Measure with CloudWatch, then tune.
Last updated: June 2026 · Written by Morten Andersen and reviewed by Fredrik Filipsson · See, Cut, Lock, Run method
Aurora Serverless v2 is the on-demand, auto-scaling configuration of Amazon Aurora that adjusts database capacity in fine-grained steps to match load, billing only for the capacity in use. It scales in Aurora Capacity Units, and you control its cost through two numbers: the minimum ACU, which is the floor you pay for at all times, and the maximum ACU, which caps how high it can scale. Get these two bounds right and Serverless v2 is genuinely economical; get the minimum wrong and you pay around the clock for headroom you do not use. This guide is part of our complete guide to AWS cost optimization, the cluster pillar it links up to.
What is an ACU and how does it bill?
An ACU, or Aurora Capacity Unit, is the unit Aurora Serverless v2 scales and bills in, and each ACU provides roughly 2 GiB of memory with proportional CPU and networking. Aurora scales the cluster up and down by the second within the minimum-to-maximum range you set, and bills per ACU-hour for whatever capacity is active during each interval. As of mid-2026 the rate is about $0.12 per ACU-hour on Aurora Standard and about $0.156 per ACU-hour on Aurora I/O-Optimized in US East, with storage and I/O billed separately. Because the minimum is always active when the cluster is running, it is the single most important cost setting. Confirm the current rates on the Amazon Aurora pricing page, since they vary by Region.
How do I set the minimum ACU without overpaying?
Set the minimum to the lowest capacity that still keeps the database healthy, not to a comfortable round number. Because you pay for at least the minimum every hour the cluster runs, an over-set floor is the most common and most expensive Serverless v2 mistake: a minimum of 8 ACUs on a workload that idles at 1 burns capacity all night and all weekend. Read the real floor from the CloudWatch ServerlessDatabaseCapacity and ACUUtilization metrics over a representative week, and set the minimum just above the level that keeps the buffer cache warm and connections responsive. AWS explains how the range affects scaling behavior in its guide to how the ACU minimum and maximum range impacts scaling. For caching layers that often sit in front of Aurora, see our sibling guide on ElastiCache cost optimization.
The minimum is a floor you pay continuously, not a safety cushion. Every ACU of minimum you do not need is billed around the clock. When in doubt, set it lower and let Aurora scale up on demand, which it does within seconds.
Should I set the minimum to 0 and use auto-pause?
Set the minimum to 0 ACUs when the workload can tolerate a brief resume delay, because that enables automatic pause and drops compute cost to zero while idle. Aurora Serverless v2 now supports scaling to 0 ACUs: with a minimum of 0, the cluster automatically pauses after an inactivity period you choose between 5 minutes and 24 hours, and while paused there are no compute charges, though storage continues to bill. This is ideal for development and test databases, internal tools, and intermittent workloads, where a paused database might cost only a few dollars a month in storage. The trade is a short resume latency on the first connection after a pause, so reserve scale-to-zero for clusters where that pause is acceptable and keep production user-facing databases at a small non-zero floor. AWS documents the behavior in its post on scaling to 0 capacity with Aurora Serverless v2.
How do I set the maximum ACU?
Set the maximum high enough to absorb real peaks but low enough to act as a budget guardrail. The maximum does not cost anything until the workload scales into it, so its job is twofold: ensure the database can reach the capacity a genuine peak needs, and cap the damage from a runaway query or traffic spike that would otherwise scale capacity, and cost, without limit. Read the real peak from the same CloudWatch capacity metric, add modest headroom, and set the maximum there. Setting it too low throttles legitimate peaks; setting it effectively unlimited removes the cost ceiling. The balance is a ceiling a little above your true peak.
Paying for Aurora capacity that sits idle overnight?
Our AWS cost audit reads your ACU utilization, sets the minimum to the true floor or to zero where auto-pause fits, and caps the maximum as a guardrail, proven against a clean baseline. On the performance model, you pay only from realized savings. No savings, no fee.
Book a cloud cost audit →How do I tune ACU min and max, step by step?
- Measure real capacity usage. Review ServerlessDatabaseCapacity and ACUUtilization in CloudWatch over a representative period. Expected result: the true floor and peak.
- Set the minimum to the true floor. Use the lowest healthy capacity, or 0 for dev and intermittent clusters. Expected result: no paying for unused floor.
- Set the maximum as a cost ceiling. Just above the real peak, as a budget guardrail. Expected result: peaks absorbed, runaway spend capped.
- Configure the auto-pause idle period. When the minimum is 0, choose 5 minutes to 24 hours of inactivity. Expected result: idle clusters pause and stop billing compute.
- Verify cost and scaling. Confirm scale-down in quiet periods and no surprise maximum hits. Expected result: a confirmed, durable reduction.
The AWS Cost Optimization Field Guide includes the Aurora Serverless v2 capacity-tuning worksheet we use to set the minimum, maximum, and auto-pause window from real metrics.
Frequently asked questions
What is an ACU in Aurora Serverless v2?
An ACU, or Aurora Capacity Unit, is the unit Aurora Serverless v2 scales and bills in. Each ACU provides about 2 GiB of memory with proportional CPU and networking. You set a minimum and maximum ACU range, and Aurora scales within it by the second based on load, billing per ACU-hour.
How much does an Aurora Serverless v2 ACU cost?
As of mid-2026, Aurora Serverless v2 costs about $0.12 per ACU-hour on Aurora Standard and about $0.156 per ACU-hour on Aurora I/O-Optimized, in US East. Storage and I/O are billed separately. Verify current rates on the Amazon Aurora pricing page, as they vary by Region.
Can Aurora Serverless v2 scale to zero?
Yes. Setting the minimum capacity to 0 ACUs enables automatic pause after an idle period you choose, between 5 minutes and 24 hours. While paused there are no compute charges, though storage continues to bill. This suits development, test, and intermittent databases; the next connection triggers a brief resume.
What is the most common Aurora Serverless v2 cost mistake?
Setting the minimum ACU too high. Because you are billed for at least the minimum at all times, an over-set floor means paying for capacity the workload does not use during quiet hours. Set the minimum to the true floor, or to 0 where auto-pause fits, rather than a comfortable round number.
The short version
Aurora Serverless v2 cost is governed by the ACU range. You pay for the minimum continuously, so set it to the real floor, or to 0 for development and intermittent clusters to get auto-pause. Set the maximum just above your true peak as a budget guardrail. Read both from CloudWatch, remember storage bills even while paused, and re-check after a week. When you want the ACU bounds set from real metrics and the Aurora bill proven down, that is part of what our AWS cost optimization service delivers.
Cloud pricing and service behavior change frequently. Verify the specifics in this guide against the providers’ own current documentation and the FinOps Foundation: AWS pricing ↗, AWS documentation ↗ and FinOps Foundation Framework ↗. This article also reflects Cloud Cost Room’s hands-on, vendor-neutral engagement experience.