Learn the tools, techniques, and strategies to maximize your AWS landscape while minimizing the cost.

'AWS Cost optimization' Course Cover Art.png

Your organization weighed the options, ran the numbers, took a deep breath, and migrated to the cloud. That’s a perfectly solid strategic move—but your monthly invoices aren’t quite as lean as your original business case predicted. What’s worse, they are actually larger than your prior costs and increasing! This, unfortunately, is not uncommon. However, you can manage your resources with smart cost optimization techniques to shrink those bills. 

Optimizing your costs–regardless of the cloud provider–takes some work. In this course, you'll discover methods and tools on AWS to optimize cloud costs, including best practices for efficiency, cost transparency, and automation. 

You'll gain a comprehensive grasp of cloud economics, cost considerations, and a practical toolbox for streamlining AWS workloads with reduced manual effort. By implementing some strategies in this course, you will significantly reduce the chances of ever getting that surprise AWS invoice again!

AWS Cost Optimization

After one month of running your workload on AWS, you notice that a DEV environment code repository system running on a single `m6i.2xlarge` EC2 instance has barely reached over 20% utilization. What would you recommend to the team to lower the costs?

After one month of running your workload on AWS, you notice that a DEV environment code repository system running on a single <code>m6i.2xlarge</code> EC2 instance has barely reached over 20% utilization. What would you recommend to the team to lower the costs?

No, because COBOL is not one of the languages which Lambda supports currently.


No, because serverless is not an appropriate design in this case.

Maybe, because Lambda can run COBOL using a custom runtime container, but there may be other data and resources used within that business logic which would not be accessible.

Yes, because we can port the COBOL code into Rust without much trouble and expose the business logic using API Gateway and REST calls.

You have been asked to consult with a company migrating their aging AS/400 workloads to the AWS. They have some very industry-specific and proprietary business logic written in COBOL. That business logic has been functioning perfectly for decades, although the original developers have long since retired. The company is in a rush because their current data center lease is expiring in a month, leaving no time for rearchitecting or porting this business logic to a more modern language. One of the company’s junior engineers asks if Lambda could host the business logic. What is your response?

You have been asked to consult with a company migrating their aging AS/400 workloads to the AWS. They have some very industry-specific and proprietary business logic written in COBOL. That business logic has been functioning perfectly for decades, although the original developers have long since retired. The company is in a rush because their current data center lease is expiring in a month, leaving no time for rearchitecting or porting this business logic to a more modern language. One of the company’s junior engineers asks if Lambda could host the business logic. What is your response?

What would be the most appropriate and cost-effective option for orchestrating a complex data conversion workflow on AWS?

What would be the most appropriate and cost-effective option for orchestrating a complex data conversion workflow on AWS?

Receive uploaded invoices into an S3 bucket. Use a fleet of EC2 Auto Scaling P4 spot instances to run the invoice through metadata extraction. Store the results in an RDS-based MySQL database.

Receive the uploaded invoices into an EFS share. Use ECS containers to perform the metadata extraction using a custom SageMaker container. Store the results in an SQS queue, which is then periodically polled via EventBridge and uploaded to an EBS data store.

Receive the uploaded invoices via API Gateway and pass them to SNS. Use AWS Glue to poll the SNS queue and hand them off to a custom SageMaker endpoint. Store the metadata in DynamoDB, provisioned as On-Demand.

Receive the uploaded invoices into an S3 bucket. Use an S3 PUT trigger to launch a lambda function that submits the invoice to Amazon Textract. When the metadata is returned, use the AWS SDK to attach that metadata to the S3 file.

You are responsible for designing an architecture for an online document processing service. Customers upload images of invoices and then metadata is extracted and associated with the original file. Which architecture would be the most cost-effective?

You are responsible for designing an architecture for an online document processing service. Customers upload images of invoices and then metadata is extracted and associated with the original file. Which architecture would be the most cost-effective?

They can cost as much as 70% less than the same instance bought on demand.

They cannot be used for workloads where data can be ephemeral and not persisted.

Spot Instances are not always available in all regions and AZs.

The Spot Instance price for an instance type can in some rare cases be pricier than On-Demand.

Spot Instances come with a service-level agreement ensuring they will stay up or your money back.

**(Select all that apply.)** Which statement is false about EC2 Spot Instances? 

(Select all that apply.) Which statement is false about EC2 Spot Instances?

Test your knowledge of the topics covered in this chapter.

Introduction

The Economics of Cloud Computing

The AWS Cost Optimization Pillar

Rightsizing

Elasticity

Pick the Right Pricing Model

Measure, Monitor, Improve

More Ways to Save

Wrap Up and Keeping Current

You Make The (Cost Optimization) Call!

Quiz: Elasticity