New – Rightsizing Recommendations for Amazon RDS MySQL and RDS PostgreSQL in AWS Compute Optimizer

Starting today, AWS Compute Optimizer delivers new recommendations for Amazon Relational Database Service (RDS) MySQL and PostgreSQL databases. These recommendations help you detect idle RDS instances and identify the optimal instance types and provisioned IOPS settings for your existing RDS DB instances, so you can reduce costs for idle and over-provisioned RDS DB instances or increase the performance of under-provisioned workloads.

Amazon RDS, a managed service, allows you to set up, operate, and scale a relational database, while automating time-consuming database admin tasks, such as, hardware provisioning, patching, and backups. While you can easily scale up your RDS database instances to handle varying business needs, it’s also important to optimize your cost as you scale. You can take advantage of RDS Reserved Instances pricing and leverage RI purchase recommendations in AWS Cost Explorer. You can also turn off RDS DB instances that are not being used, avoid over-provisioning for your applications, and upgrade to the latest generation instances for better price-performance.

However, getting the insights to help you optimize can be time-consuming. You need to dedicate engineering resources specialized in databases to analyze the cost and performance or iteratively adjust resources to achieve the optimal configuration. With these new recommendations in AWS Compute Optimizer, you can identify opportunities to optimize your database resource configurations faster.

Introducing Rightsizing Recommendations for Amazon RDS

A recommendation for Amazon RDS MySQL and RDS PostgreSQL databases includes two parts. First, you will see a recommendation for the DB instance, and second, you will see a recommendation for the DB instance storage.

Figure 1. Sample screenshot of Amazon RDS rightsizing recommendations in AWS Compute Optimizer

The DB instance part of the recommendation tells you whether Compute Optimizer thinks the database might be idle. Then it gives you up to two instance recommendation options (one x86 and one Graviton) based on whether Compute Optimizer finds the database to be optimized, under-provisioned, or over-provisioned. You can then compare options with the current DB instance to understand the cost and specification differences. Similarly, in the storage tab, you will find a recommendation for the storage type (such as upgrades to gp3 from gp2) and provisioned IOPS for io1 storage volumes.

Wabtec has been able to reinvigorate our RDS optimization initiatives by leveraging the insights and recommendations provided by AWS Compute Optimizer for RDS, which has streamlined engineers’ ability to analyze and act upon RDS optimization opportunities. With single-click access to data, we quickly optimized 40% of the pilot group. This data-driven approach has accelerated a comprehensive roadmap for adopting AWS Graviton instances, which was backed by Compute Optimizer cost savings forecasting.

David Poulliott
Director, Cloud & Application Architecture | Tech Services | Wabtec Corporation

To generate these recommendations, Compute Optimizer analyzes metrics from Amazon CloudWatch such as CPU utilization, database connections, network throughput, and storage IOPS and throughput. If you have Amazon RDS Performance Insights enabled on your RDS DB instance, Compute Optimizer also analyzes metrics, such as DBLoad, swap in, and swap out to give you even more insights. Each of these utilization metrics are presented in the same view to help you evaluate each recommendation option. You can even adjust the lookback period to 32-days or 93-days using rightsizing preferences, so that Compute Optimizer can accommodate cyclical patterns in your usage and generate a recommendation more fitted to your workload.

Figure 2. Sample screenshot of RDS utilization metrics view in AWS Compute Optimizer

“We use a lot of RDS instances for our business and are always looking for ways to analyze and optimize the performance and utilization of our RDS workloads. Thanks to Amazon RDS Rightsizing Recommendations, we were able to more effectively detect idle databases as well as increase our efficiency by changing to different instance types. Our engineering teams are now more confident in implementing change due to the depth of included data.”

Dave Van Hoven
Senior DevOps Manager | Cloud & Employee Foundations | Here Technologies

Accessing the recommendations

If you have already opted into Compute Optimizer, rightsizing recommendations for RDS are automatically enabled. New customers who have not opted in Compute Optimizer can enable the service with a few clicks in the Compute Optimizer Console or through the Compute Optimizer API. Once activated, Compute Optimizer needs at least 30 hours of metrics before it delivers recommendations for an RDS DB instance. For more details, see user guide “Getting started with AWS Compute Optimizer”.

All your recommendations across accounts, regions, and resource types in one place

If you use Cost Optimization Hub to aggregate and prioritize all of your organization’s savings opportunities, you will automatically see these new recommendations for Amazon RDS. The recommendations are presented as two resource types – RDS DB instance, and RDS DB instance storage. Here you will see recommended actions to: stop idle or unused databases, rightsize DB instance types and storage volumes, upgrade to a later-generation product, migrate to Graviton, and purchase Reserved Instances for Amazon RDS.

Figure 3. Sample screenshot of RDS rightsizing recommendations in Cost Optimization Hub

Conclusions

With these new recommendations in Compute Optimizer, you can quickly detect idle RDS resources and identify opportunities to optimize your RDS DB instance and storage usage, so you can balance the costs and performance for each of your workloads. In Cost Optimization Hub, you get a clear understanding of the potential savings for your RDS workloads and evaluate and prioritize them among all of your other savings opportunities across your organization.