Compute Optimizer recommends resource configurations and identifies idle resources.
Resource rightsizing
AWS Compute Optimizer helps choose resource configurations based on usage. RightsizingChoosing a resource configuration suited to the workload's capacity and performance needs. means selecting capacity suited to the workload rather than provisioning too much or too little.
How it works
- Opt in an account. An organization management account can opt in all member accounts, including members added later.
- Compute Optimizer examines resource configuration and utilization metricsA measurement of how much of a resource is being used., measurements of how resources are used. Supported examples include EC2 instances and Auto Scaling groups, EBS volumes, Lambda functions, ECS services on Fargate, and Aurora and RDS databases.
- Eligible resources need enough data. Analysis can take up to 24 hours once sufficient metrics are available.
- Review recommendations alongside recent and projected utilization graphs. Adjust rightsizing preferences such as CPU and memory headroomCapacity reserved above observed utilization., the spare capacity reserved above observed use, when appropriate.
Exam cues
- If the question says find a better EC2 size based on usage, think Compute Optimizer, because it analyzes configuration and utilization.
- If the question says analyze a longer usage history, think enhanced infrastructure metrics, because the paid preference extends the lookbackThe historical period of resource metrics used for analysis. for selected resources.
- If the question says enforce software license consumption rules, think License Manager, because it tracks licenses and provides rule-based controls.
Remember
- Default lookback: 14 days. Rightsizing preferences also offer 32 days without additional payment, or 93 days with paid enhanced infrastructure metrics for selected resources.
- For EC2 memory metrics, use the CloudWatch agent or supported external metrics ingestion.
- Enable Cost Explorer for pricing and savings calculations using billing data.
- Commercial software license recommendations cover Microsoft SQL Server on EC2.
Trap: Compute Optimizer recommends configurations, but License Manager provides rule-based license consumption controls.
Da documentação da AWS (13)
- Compute Optimizer analyzes resource configuration and utilization metrics to recommend sizes and identify idle resources.
- Compute Optimizer requires an account to opt in before it analyzes resources.
- An organization management account can opt in all member accounts, including new members added later.
- Supported resources include EC2 instances and Auto Scaling groups, EBS volumes, Lambda functions, ECS services on Fargate, and Aurora and RDS databases.
- The default analysis lookback is 14 days; paid enhanced infrastructure metrics extend it to 93 days for selected resources.
- Rightsizing preferences offer 14-day and 32-day lookbacks without additional payments and a paid 93-day option.
- After sufficient metrics are available, analysis can take up to 24 hours.
- Compute Optimizer provides recent and projected utilization graphs to help evaluate recommendation trade-offs.
- Rightsizing preferences can adjust CPU threshold, CPU and memory headroom, lookback periods, and instance family preferences; CPU and memory utilization preferences apply only to EC2 instances.
- Compute Optimizer can use EC2 memory utilization collected by the CloudWatch agent or ingested from supported external products.
- Cost Explorer must be enabled for Compute Optimizer to calculate savings and populate pricing information for recommendations using billing data.
- AWS License Manager tracks license usage and provides rule-based controls on license consumption.
- Compute Optimizer generates commercial software license recommendations only for Microsoft SQL Server on EC2.
Recommendation eligibility
AWS Compute Optimizer recommendations depend on resource-specific requirements. When a recommendation is missing, check the requirements for that resource type.
How it works
- For EC2 instances and Auto Scaling groups, check for at least 30 hours of CloudWatch metrics in the past 14 days, or in the past 93 days with enhanced infrastructure metrics.
- For Lambda memory recommendations, check both configured memory and invocation count in the last 14 days.
- For a supported EBS volume, check that it has been attached to a running instance for at least 30 consecutive hours. Its recommendation is unavailable after detachment.
Exam cues
- If the question says a new EC2 instance has only 10 hours of metrics, think insufficient data, because it needs at least 30 hours in the applicable lookback periodThe historical period of resource metrics used for analysis..
- If the question says a Lambda function has only 20 invocations in 14 days, think insufficient invocations, because memory recommendations require at least 50.
- If the question says an EBS recommendation disappeared after detachment, think attachment state, because recommendations are unavailable for detached volumes.
Remember
- EC2 instances and Auto Scaling groups need 30 hours of CloudWatch metrics in 14 days, or 93 days with enhanced infrastructure metrics.
- Lambda memory recommendations require at least 50 invocations in the last 14 days and configured memory of at most 1792 MB.
- Supported EBS volumes need 30 consecutive hours attached to a running instance.
Trap: The 30-hour EC2 metric requirement and the 30-consecutive-hour EBS attachment requirement are different checks.
Da documentação da AWS (3)
- EC2 instances and Auto Scaling groups require at least 30 hours of CloudWatch metrics in the past 14 days, or 93 days with enhanced infrastructure metrics.
- Lambda memory recommendations require configured memory of at most 1792 MB and at least 50 invocations in the last 14 days.
- Supported EBS volumes must be attached to a running instance for at least 30 consecutive hours; recommendations disappear when a volume is detached.
Flashcards
Nada para revisar aqui agora. Escolha outros tópicos ou outro filtro.
·
Todos os 8 flashcards em lista
Which AWS service recommends resource sizes based on configuration and utilization?
AWS Compute Optimizer analyzes resource configuration and utilization to recommend sizes and identify idle resources.
What is the default AWS Compute Optimizer lookback, and how can selected resources use a longer period?
The default is 14 days. Paid enhanced infrastructure metrics extend it to 93 days for selected resources.
A team needs to find idle resources and choose better EC2 instance sizes based on usage. Which AWS service can help?
AWS Compute Optimizer analyzes configuration and utilization to identify idle resources and recommend sizes.
A team is evaluating an AWS Compute Optimizer rightsizing recommendation. What can it use to assess the utilization trade-off?
Review the recent and projected utilization graphs provided by Compute Optimizer.
A team wants AWS Compute Optimizer EC2 recommendations to consider instance memory use. What can it configure?
Collect memory metrics with the CloudWatch agent or ingest them from a supported external product.
A team sees no AWS Compute Optimizer recommendation for a new EC2 instance with only 10 hours of metrics. Why?
EC2 recommendations require at least 30 hours of CloudWatch metrics in the applicable lookback period.
A Lambda function has 20 invocations in 14 days. Why does AWS Compute Optimizer not recommend its memory size?
Lambda memory recommendations require at least 50 invocations in the last 14 days and configured memory of at most 1792 MB.
A team detached an EBS volume and its AWS Compute Optimizer recommendation disappeared. Why?
Recommendations are unavailable when a volume is detached; supported volumes need at least 30 consecutive hours attached to a running instance.