Aurora Serverless v2 0 Capacity Scaling

Question

I was just notified that my Aurora Serverless v1 clusters will be auto-upgraded to v2 in about one year.

**Background**

I am currently using Serverless v1 for an open source project, [cdk-static-wordpress](https://github.com/blimmer/cdk-static-wordpress), which hosts an ephermal Wordpress admin instance that produces static content served by S3 + CloudFront. So, you just need to spin up the Wordpress admin instance when you need to mutate content. Then, when you're done, you generate the static site and spin down the admin instance.

The goal of this project is to host Wordpress as cheaply as possible, only paying for the ECS task and Database when running the admin instance.

**Serverless v1 vs. v2**

To reduce the cost of the admin instance, I'm using Serverless v1 because it can scale down to 0 capacity when the admin instance is not running. However, according to the documentation, Serverless v2 has a minimum capacity of 0.5 units. In `us-west-2`, 0.5 ACU for MySQL is $0.12/hour. So upgrading from v1 to v2 will be prohibitively expensive for this project's goals.

**Main Question**

How are other people in this situation (using Serverless v1, relying on scaling to 0) planning to migrate off of Serverless v1?

**Alternative Options**

A few alternative options I've thought about:

- Stopping the DB. I know you can stop the DB for up to 7 days, but it will be auto-started again after 7 days. So, this doesn't feel like a great solution.
- RDS/EC2 on free tier. This works for folks eligible for free tier, but not everyone can use free tier.
- Run database as an additional ECS service with EFS backing.

Answer

Hello,

Thanks for writing to us via this forum.

In these kind of situations usually we recommend customers to be

1. Aware of the architectural configurations (which you are already well versed on) - as you correctly stated the differences in ACU for v1 and v2.
2. Take necessary actions as per the architecture. So we see customers either switching platforms (v1 to v2 or vice versa or go to provisioned or RDS) or modify the use case for the cluster (send a different type of workload to the cluster that can play favorably with the configurations the cluster has to offer)

We cannot generalize what customers usually do as it all depends on per customer basis and every customer has their own process and workload to manage. Hope that helps with your question.

Aurora Serverless v2 0 Capacity Scaling

Relevant content