Cost visibility and rightsizing: SAA-C03 study notes and practice questions
What to know about Cost visibility and rightsizing for the SAA-C03 exam, from the Design Cost-Optimized Architectures domain (20% of the exam guide), then 2 free questions to test it. Part of the SAA-C03 study guide.
Cost management questions ask how to see, control or reduce spending with as little effort as possible. Each tool has a specific job.
The tools
Tool
Use it to
AWS Cost Explorer
View and filter costs and usage over time, forecast, and see Savings Plans and RI recommendations
AWS Budgets
Set a cost, usage or coverage budget and alert by email or SNS when actual or forecasted spend crosses a threshold; budget actions can apply an IAM or SCP policy or stop instances
AWS Cost Anomaly Detection
Use machine learning to spot unusual spend and alert the owner, without setting thresholds
AWS Compute Optimizer
Recommend rightsizing for EC2, Auto Scaling groups, EBS, Lambda, ECS on Fargate and RDS, from utilization history (14 days by default; memory needs the CloudWatch agent)
AWS Trusted Advisor
Check for idle resources, underused instances and other savings (full checks need Business Support or higher)
Cost and Usage Reports / Data Exports
Detailed line items to S3, queried with Athena or QuickSight
Cost allocation tags
Attribute costs to teams or projects; activate them in the Billing console
AWS Cost Optimization Hub
Gather savings recommendations across accounts in one place
Key facts
Tags must be activated as cost allocation tags before they show up in billing tools, and they only apply from activation onward.
Tag policies in AWS Organizations standardize tag keys and values across accounts.
Budgets alert; they don’t stop spending unless you configure a budget action.
Consolidated billing gives one bill and shares volume discounts and commitments across the organization.
How to choose
“Warn the team before the monthly budget is exceeded”: Budgets with a forecasted alert.
“Find out which team caused this month’s increase”: cost allocation tags with Cost Explorer.
“Catch unexpected spikes without setting thresholds”: Cost Anomaly Detection.
“Instances look oversized; recommend new sizes with no custom tooling”: Compute Optimizer.
Common traps
CloudWatch billing alarms for forecasts. They work on actual estimated charges; Budgets can alert on forecasts.
Expecting historic data for a newly activated tag.
Rightsizing on CPU alone when memory is the bottleneck. Compute Optimizer needs memory metrics from the agent.
A company runs 200 Amazon EC2 instances that host in-memory Java services.
The finance team suspects that many instances are oversized, but engineers warn that the services are limited by memory, not CPU, and that resizing based on CPU data alone has caused outages before. The company wants instance type recommendations based on several weeks of real usage, and it does not want to build its own analysis tooling.
Which solution meets these requirements?
Choose one answer.
AA. Opt in to AWS Compute Optimizer, and use its default analysis of the metrics that Amazon EC2 already sends to Amazon CloudWatch.
BB. Opt in to AWS Compute Optimizer, and activate enhanced infrastructure metrics so that recommendations use up to 93 days of history.
CC. Install the CloudWatch agent, and write an AWS Lambda function that compares each instance's memory use with instance type specifications.
DD. Opt in to AWS Compute Optimizer, and install the Amazon CloudWatch agent on the instances to publish memory utilization metrics.
Show the answer and explanation
Answer: D
A. Wrong: Compute Optimizer is the right service for right-sizing recommendations, but EC2 does not report memory use to CloudWatch by itself. Without memory data, recommendations rest on CPU, network and disk metrics, which is the CPU-only resizing the engineers warned about. Learn more: What is AWS Compute Optimizer? (opens in a new tab)
B. Wrong: Enhanced infrastructure metrics is a paid feature that extends the lookback period from 14 days to up to 93 days. A longer history does not add memory data, so the recommendations would still be blind to the services' real limit. Memory needs the CloudWatch agent. Learn more: Enhanced infrastructure metrics (opens in a new tab)
C. Wrong: This would capture memory use, but the company would have to build and maintain the comparison logic, pricing data and reports itself. That is the in-house analysis tooling it wants to avoid, when Compute Optimizer already does this analysis. Learn more: EC2 instance metrics analyzed by Compute Optimizer (opens in a new tab)
D. Right: Correct. Compute Optimizer analyses each instance's utilization history and recommends better-fitting instance types. Once the CloudWatch agent publishes memory utilization with the InstanceId dimension, Compute Optimizer includes memory in its analysis, so recommendations reflect the real constraint. Learn more: EC2 instance metrics analyzed by Compute Optimizer (opens in a new tab)
Explanation
AWS Compute Optimizer gives right-sizing recommendations without custom tooling, and by default it looks at 14 days of history, which covers several weeks of real use. It can only consider memory if memory metrics exist, and EC2 does not publish them by default, so the CloudWatch agent must be installed. Extending the lookback period does not fix the missing memory data.
A small start-up has set itself a limit of 500 US dollars a month for AWS spending.
The founders want an email warning partway through the month if spending is on track to go over this amount, so they can act before the bill arrives. They do not want to write or maintain any code.
Which solution meets these requirements?
Choose one answer.
AA. Create a monthly cost budget of 500 dollars in AWS Budgets, with an email alert when actual spend exceeds the budget.
BB. Turn on AWS Cost Anomaly Detection with a monitor for all AWS services and a daily email alert subscription.
CC. Create a monthly cost budget of 500 dollars in AWS Budgets, with an email alert when forecasted spend exceeds the budget.
DD. Write an AWS Lambda function that calls the Cost Explorer API each day and emails the founders through Amazon SNS.
Show the answer and explanation
Answer: C
A. Wrong: AWS Budgets is the right tool and needs no code, but an alert on actual spend only fires once the money has already been spent. The founders want warning while there is still time to act, which needs an alert on forecasted spend. Learn more: Managing your costs with AWS Budgets (opens in a new tab)
C. Right: Correct. AWS Budgets can alert on forecasted spend as well as actual spend, and it sends notifications by email or Amazon SNS. A forecast alert warns the founders during the month, before the limit is reached, with nothing to build. Learn more: Managing your costs with AWS Budgets (opens in a new tab)
D. Wrong: A custom function could work out the month's spending and send a warning, but it is code that the start-up would have to write and maintain. AWS Budgets already provides forecast alerts without any code.
Explanation
AWS Budgets tracks spending against an amount you set and can alert on forecasted spend, so it warns before the limit is reached rather than after. Alerts go by email or Amazon SNS and need no code. Anomaly detection finds unusual patterns, not a fixed limit, and a custom function adds code to maintain.
Want more on Cost visibility and rightsizing? The 5 full-length practice sets have 10 more Cost visibility and rightsizing questions, 325 in all, none of them in the free exam. One payment of $14.99, no subscription. See the practice sets →