AWS Compute Services

• AWS offers a wide array of compute services

AWS Compute Services

• Amazon EC2 is a virtual server in the cloud
• Amazon EC2 Auto Scaling automatically adjusts the number of EC2 instances based on demand
• Amazon Elastic Container Registry (ECR) is a fully-managed Docker container registry
• Amazon Elastic Container Service (ECS) is a container management service
• VMware Cloud on AWS allows running VMware workloads on AWS infrastructure
• AWS Elastic Beanstalk provides an easy way to deploy and manage web applications
• AWS Lambda allows running code without provisioning or managing servers
• Amazon Elastic Kubernetes Service (EKS) is a managed Kubernetes service
• Amazon Lightsail offers easy-to-use virtual private servers (VPS) with simplified management
• AWS Batch enables running batch computing workloads
• AWS Fargate is a serverless compute engine for containers
• AWS Outposts brings AWS infrastructure and services to on-premises locations
• AWS Serverless Application Repository allows discovering and deploying serverless applications

Categorizing Compute Services

• Amazon EC2 is categorized as Infrastructure as a Service (IaaS)
• Amazon EC2 is instance-based, and gives access to virtual machines
• AWS Lambda is a serverless compute service that is function based
• AWS Lambda is low cost
• Amazon ECS, Amazon EKS, AWS Fargate, and Amazon ECR are all examples of container-based computing and are instance-based
• AWS Elastic Beanstalk is categorized as Platform as a Service (PaaS)
• AWS Elastic Beanstalk is optimized for web applications

Choosing the Optimal Compute Service

• Selecting the right compute service depends on the use case
• Consider application design, usage patterns, and configuration settings to manage
• Selecting the wrong solution can reduce performance efficiency
• Understanding available compute options is a good starting point

Amazon EC2 Pricing Models

• On-Demand Instances allow paying by the hour without long-term commitments, and are elegible for the AWS Free Tier
• Reserved Instances offer discounted hourly rates with a 1-year or 3-year term, and require upfront payment
• Spot Instances runs as long as available and the bid is above the Spot Instance price, and can be interrupted by AWS with a 2-minute notification
• Dedicated Hosts are physical servers with EC2 capacity, fully dedicated to your use
• Dedicated Instances run in a VPC on hardware dedicated to a single customer
• Per-second billing available for On-Demand, Reserved, and Spot Instances
• Spot Instances run Amazon Linux or Ubuntu
• On-Demand Instances provide low cost and flexibility
• Spot Instances allow for large scale dynamic workloads
• Reserved Instances provide predictable compute capacity
• Dedicated Hosts helps to save money on licensing costs and meet compliance/regulatory requirements

Amazon EC2 Pricing Use Cases

• On-Demand Instances are best for short-term, spiky, unpredictable workloads and application development/testing
• Spot Instances suits applications with flexible start/end times and urgent computing needs with additional capacity
• Reserved Instances are used for steady state or predictable workloads with reserved capacity including disaster recovery
• Dedicated Hosts suits compliance/regulatory restrictions and helps to bring your own license (BROL)

Four Pillars of Cost Optimization

• Pillar 1 is to right-size by provisioning instances to match demand, and right sizing after reserve
• Pillar 2 is to increase elasticity by stopping/hibernating instances that are not in use and using automatic scaling for elasticity
• Pillar 3 is to consider the right pricing model for the use case, optimize purchase types and use serverless solutions
• Pillar 4 is to reduce costs and increase storage by resizing EBS volumes, use Storage Throughput Optimized HDD, resize or delete unneeded files

Measure, Monitor and Improve

• Cost optimization is ongoing and must be continuously improved
• Cost allocation tagging must be defined and enforced

Container Basics

• Containers are a method of OS virtualization
• Containers are repeatable, self-contained, and run the same in different environments
• Containers offer faster launch and termination than virtual machines

Docker

• Docker enables building, testing, and deploying applications quickly
• Running containers on Docker makes them from a template called an image
• A container includes everything a software app needs to run

EC2

• EC2 is an example of Infrastructure as a Service (IaaS)

Amazon Elastic Container Service (ECS)

• Amazon ECS facilitates running Docker containers in a managed cluster
• Amazon ECS removes complexity of standing up the infrastructure, by orchestrating running Docker containers, and maintains and scales the fleet of nodes that run your containers
• Elastic Load Balancing, Amazon EC2 security groups, Amazon EBS volumes, and IM roles are integrated
• Create an Amazon ECS cluster backed by Amazon EC2 when granular control is needed
• Create an Amazon ECS cluster backed by AWS Fargate to maintain easier, and focus on applications

Kubernetes

• Kubernetes is open source software for container orchestration
• Kubernetes enables to deploy and manage containerized applications at scale
• Kubernetes helps to run multiple containers on a single OS host
• Kubernetes automates container provisioning, networking, load distribution, and scaling

Amazon Elastic Kubernetes Service (EKS)

• Amazon EKS enables running Kubernetes on AWS
• Amazon EKS is certified and supports Linux and Windows containers
• Amazon EKS helps to manage clusters of Amazon EC2 compute instances and run containers

Amazon Elastic Container Registry (ECR)

• Amazon ECR is a fully managed Docker container registry
• Amazon ECR integration, Docker support, team collaboration, access control, and third-party integrations
• Amazon ECR makes it easy to store, manage, and deploy Docker container images

AWS Lambda

• AWS Lambda is a serverless compute service
• AWS Lambda supports multiple programming languages, completely automated administration, and built-in fault tolerance
• AWS Lambda supports the orchestration of multiple functions with pay-per-use pricing
• The code runs only when it is triggered in response to events, on a schedule from AWS services that can be set up as event sources

AWS Lambda Event Sources

• The code runs only when triggered with logging, and monitoring
• Invoking a Lambda function from the Lambda console, AWS SDK, or AWS CLI as the alternative

AWS Lambda Use Case

• Lambda can be configured to automatically stop or start EC2 instances

AWS Lambda Quotas

• Concurrent executions' soft limit is 1,000
• Function and layer storage soft limit is 75 GB
• Maximum function memory allocation hard limit is 10,240 MB
• Function timeout hard limit is 15 minutes
• Deployment package size hard limit is 250 MB (unzipped)
• Container image code package size hard limit is 10 GB

AWS Elastic Beanstalk

• AWS Elastic Beanstalk is an easier way to get web applications running
• There is no additional charge for managing Elastic Beanstalk with pay only for the underlying resources
• AWS Elastic Beanstalk automatically handles infrastructure provisioning/configuration, deployment, and load balancing

AWS Elastic Beanstalk Deployments

• Web applications are supported with Java,.NET, PHP, Node.js, Python, Ruby, Go, and Docker
• AWS manages the language interpreter, and operating systems
• AWS Elastic Beanstalk automatically handles these deployments
• AWS Elastic Beanstalk lets you deploy on servers such as Apache, NGINX, Passenger, Puma, and Microsoft Internet Information Services (IIS)
• AWS Elastic Beanstalk is simple and fast to start using, helps to gain developer productivity and can be easily outgrown