AWS is introducing a centralised “Cost Optimization Hub” and new customisation options in its Compute Optimizer – a service launched in 2019, but criticised for its limitations including being restricted to just four AWS services and having a mere 14-day “lookback” period of utilisation history, which it uses to generate cost-saving recommendations.
(Compute Optimizer supports four types of AWS resources: EC2, EBS volumes, ECS services on AWS Fargate, and AWS Lambda functions.)
Today (November 26) as delegates continued to arrive in Las Vegas for the hyperscaler’s annual re:Invent conference, AWS said it was fleshing out Compute Optimizer, with the ability to set a 32-day lookback period option, and set wider instance family preferences – as well as rolling out the ability to adjust CPU headroom and thresholds. These new settings, it said, can be configured at the organisation, account, or regional level.
That longer lookback period lets organisations capture data like monthly patching, reboots, and maintenance activities when seeking to optimise.
Join peers following The Stack on LinkedIn
Compute Optimizer also offers the option of a 93-day lookback period with “Enhanced Infrastructure Metrics” at ~$0.25/month per instance.
“Compute Optimizer supports 651 instance types today and, by default, evaluates all available instances to identify the top three options with the least performance risk and the most cost savings based on your utilization history” AWS product manager Wenyen Huang said today in a blog.
He added that “when it comes to rightsizing, you may have organizational guidelines or requirements regarding instance types. For example, you may have purchased Amazon EC2 Instance Savings Plans and Reserved Instances and only wish to rightsize to instance families covered by those commitments. You may want to use only instances equipped with certain processors or non-burstable instances due to your application design. Additionally, the workloads you are running may require instance types certified by Independent Software Vendors (ISVs), such as SAP HANA workloads. This feature allows you to define a custom instance type list that compute optimizer will use to create its recommendations.”
AWS Cost Optimizer: What else is new?
Further new capabilities in Compute Optimizer include the ability to tweak utilisation threshold in order to remove transient outlier utilisation spikes from the recommendation data – scaling a workload based on these can lead to over-provisioning costs. Compute Optimizer now offers three utilisation threshold options (P90, P95, P99.5) for CPU, said Huang.
By default, Compute Optimizer uses a P99.5 threshold, which means it ignores the top 0.5% of the highest utilisation data points from your history. If you set this to P90 instead, it ignores the top 10% of your highest data points from your utilisation history; possibly, for example, useful for non-production workloads less sensitive to peak utilisation.
New AWS Cost Optimization Hub: More in one place
A new AWS Cost Optimization Hub, meanwhile, allows customers to more easily identify, filter, and aggregate what the hyperscaler said was 15 types of AWS cost optimisation recommendations, such as EC2 instance rightsising recommendations, Graviton migration recommendations, idle resource recommendations, and Savings Plans recommendations, across customers’ AWS accounts and AWS Regions through a single dashboard.
The new AWS Cost Optimization Hub lets users spin up money-saving recommendations for the following 13 AWS services, a product page shows.
- Amazon EC2 instances
- Amazon EC2 Auto Scaling groups
- Amazon EBS volumes
- AWS Lambda functions
- Amazon ECS tasks on AWS Fargate
- Compute Savings Plans
- EC2 Instance Savings Plans
- SageMaker Savings Plans
- EC2 Reserved Instances
- Amazon RDS Reserved Instances
- OpenSearch Reserved Instances
- Amazon Redshift reserved nodes
- ElastiCache reserved nodes
Whilst the improvements will be welcomed, customers should be aware that there are a host of other potential ways to save money on cloud workloads. Airbnb saved a lot of money on AWS using Kubernetes Cluster Autoscaler intelligently and there may also be lessons there for customers.
