Services - Compute

Question 1

Batch - Characteristics

Answer 1

• It allows to run thousands of computing jobs
• It dynamically maintains the optimal compute resources like CPU, Memory based on the volume of submitted jobs
• It executes workloads on EC2 instances and Fargate
• There is no charge for using this service

Question 2

Batch - Multi-node parallel jobs

Answer 2

• Allow to run single jobs that span multiple Amazon EC2 instances
• Can run large-scale, tightly coupled, HPC applications and distributed GPU model training
• A Batch multi-node parallel job is compatible with any framework that supports IP-based, internode communication, such as Apache MXNet, TensorFlow, Caffe2, or Message Passing Interface (MPI)

Question 3

Elastic Beanstalk - Characteristics

Answer 3

• It’s service for deploying and scaling web applications and services developed with Java, .NET, PHP, Node.js, Python, Ruby, Go, Docker, and Packer builder on servers like Apache, Nginx, Passenger, and IIS
• Upload the code and it automatically handles the deployment, capacity provisioning, load balancing, auto-scaling, and application health monitoring
• Can have control over the AWS resources powering the application and access those resources at any time
• Pay for AWS services used only

Question 4

Elastic Beanstalk - Types of environments

Answer 4

• Web tier environment: handles HTTP / HTTPS requests from the users
• Worker environment:
• A worker is a background process that helps applications for handling heavy resource and time-intensive operations
• EC2 instances and SQS queues are involved

Question 5

ECS - Characteristics 1

Answer 5

• It’s a managed container orchestration service that helps to deploy, manage, and scale containerized applications
• Manages clusters of instances that support Docker containers, and monitors resource consumption and availability requirements
• Scales elastically and integrates with other services

Question 6

ECS - Characteristics 2

Answer 6

• Security is achieved through policies, availability requirements, and resource allocation
• Pay for AWS resources created to store and run the application
• When launching an ECS instance can pass user data to that instance by setting your custom configuration details in the user data parameter

Question 7

ECS - Components

Answer 7

• ECS container agent: a container that allows EC2 instances to talk with the backend logic of ECS for resource management, lifecycle coordination, and efficient scheduling
• Cluster: a group of EC2 instances spread across many AZs that are running the ECS container agent
• Task definition: a file that describes one or more containers. It’s a blueprint that specifies various parameters of your application
• Task: instantiation of a task definition within a cluster
• (Optional) Service: run and maintain many tasks simultaneously in an ECS cluster. Can optionally run it behind a load balancer

Question 8

EKS - Characteristics 1

Answer 8

• It’s a managed container service to run and scale Kubernetes applications in the cloud or on-premises
• Doesn’t need to install, operate, and maintain your own Kubernetes control plane or node
• Scales the control plane across multiple AZs, based on load

Question 9

EKS - Characteristics 2

Answer 9

• Replaces unhealthy control plane instances, and provides automated version updates and patching for them
• It’s integrated with many services such as ECR, ELB, IAM, and VPC
• Consists of EC2 instances of Fargate compute, and a control plane
• Pay for every EKS cluster used. If you run your Kubernetes worker nodes on EC2 you pay also for EC2 instances and EBS volumes used

Question 10

ELB - Characteristics

Answer 10

• It distributes incoming traffic across multiple targets, such as EC2 instances, containers, and IP addresses, in one or more AZs
• It uses health checks on its registered targets to routes traffic only to healthy targets
• It can automatically scale to the vast majority of workloads
• Cross-zone load balancing:
• When enabled, each load balancer node distributes traffic across the registered targets in all enabled AZs
• When disabled, each load balancer node distributes traffic only across the registered targets in its AZ

Question 11

ELB - Types

Answer 11

• Application Load Balancing
• Network Load Balancing: for load balancing of TCP, UDP, and TLS traffic where extreme performance is required. Routes traffic to targets within VPC and can handle millions of requests per second. Support dynamic host port mapping as ALBs do
• Classic Load Balancing: basic load balancing across many EC2 instances. Operates at request or connection level. Intended for applications that were built within the EC2-Classic network
• Gateway LB

Question 12

ELB - ALB characteristics 1

Answer 12

• Makes routing decisions at the application layer
• Supports HTTP, HTTPS, HTTP/2 and WebSockets. Can use any TCP port
• Uses round-robin load balancing by default
• Provides health checks. They consist of a HTTP GET request and a success code

Question 13

ELB - ALB characteristics 2

Answer 13

• Supports path-based routing (forward request to different target groups) and host-based routing (forward requests based on host name)
• Supports native IPv6, WAF integration, dynamic ports (ports utilized by scheduled containers of ECS), deletion protection, and request tracing
• Support Server Name Indication (SNI) to enable hosting of multiple domain names with different TLS certificates for a single ALB. Many certificates can be associated with different listeners
• Supports authentication from OIDC compliant identity providers (Google, Amazon, Facebook). It’s implemented through an authentication action on a listener rule that integrates with Cognito to create the respective user pool

Question 14

ELB - ALB components

Answer 14

• Listener: process that checks for a connection request using the configured protocol and port, and according defined rules
• Target: traffic destination based on established listener rules
• Target group: routes requests to one or more registered targets. A target can belong to multiple target groups. Health checks can be configured per target

Question 15

ELB - Gateway Load Balancing

Answer 15

• Deploys, scales, and runs third-party virtual networking appliances
• Provides load balancing and auto scaling for fleets of third-party appliances
• It’s transparent to the source and destination of traffic. It’s well suited when using third-party appliances for security, network analytics, and other use cases

Question 16

ELB - Pricing

Answer 16

• Application Load Balancer: charged for every hour of use, and the number of Load Balancer Capacity Units (LCU) used. They consist of connections, processed bytes, and rule evaluations
• Network Load Balancer: charged for every hour of use, and the number of Network Load Balancer Capacity Units (NLCU) used. They consist of connections and processed bytes
• Classic Load Balancer: charged for every hour of use and for each GB transferred through the LB
• Gateway Load Balancer: charged for every hour of use, and the number of Gateway Load Balancer Capacity Units (GLCU) used. They consist of connections and processed bytes

Question 17

Fargate - Characteristics 1

Answer 17

• It’s a managed service that you can use with ECS / EKS without having to manage servers or clusters of EC2 instances
• Pay only for the CPU and memory time used
• Customer only owns and manages their tasks, so there’s no SSH access to the infrastructure
• Compute Savings Plans can be used to reduce costs in exchange for a usage commitment

Question 18

Fargate - Characteristics 2

Answer 18

• A task defines container resources, image, and other details that indicate how the containers should run
• A task runs inside a defined VPC
• Supports ALB and NLB, not the Classic LB
• Supports advanced task-level networking, so ENIs can be assigned to running ECS tasks from VPC subnets defined by you

Question 19

Lambda - Characteristics 1

Answer 19

• It’s a serverless, event-driven compute service that lets you run code for any type of application or backend service without provisioning or managing servers
• The code is organized in Lambda functions. You pay for what you use
• Allows to add custom logic to AWS resources such as S3 buckets, DynamoDB tables, API endpoints, and others
• Every Lambda function has an IAM role associated that is specified when the function is created. Two types of permissions can be granted:
• To execute actions against other AWS resources
• To poll streams and read new records on those streams

Question 20

Lambda - Characteristics 2

Answer 20

• Allows to create backend application services that process custom events instead of servicing those events on the client
• Can use third-party libraries or native code. The languages supported are Java, Go, PowerShell, Node.js, C#, Python, and Ruby
• Provides automatic scaling. There’s no limit in the number of sequential requests
• Compute Savings Plans can be used to reduce costs in exchange for a usage commitment

Question 21

Lambda - Configuration restrictions

Answer 21

• Disk space is limited to 512 MB
• Memory allocation is limited from 128MB to 10240 MB
• Function execution is limited to 900 seconds (15 minutes)
• Restrictions exist in request and response payload, deployment package size, file descriptors, concurrent executions, and others

Question 22

Step Functions - Characteristics

Answer 22

• It’s a serverless, low-code visual workflow service used to orchestrate AWS services, like Lambda Functions. Also can automate business processes, and build applications
• Through a graphical console, you see your application’s workflow as a series of event-driven steps
• Workflows are defined using Amazon State Language

Question 23

Step Functions - Components

Answer 23

• State machine represents a workflow
• Task is type of state in a workflow. Each step in a workflow is a state

Question 24

Step Functions - Workflow types

Answer 24

• Standard workflows have exactly-once workflow execution. So a task and states are never executed more than once and can run for up to one year
• Express workflows have at-least-once workflow execution. So a task and states may be executed more than once and can run for up to five minutes

Question 25

Step Functions - State types

Answer 25

* Task: do some work in your state machine * Choice: make a choice between branches of execution * Fail / Succeed: stop an execution with a failure or success * Pass: simply pass its input to its output or inject some fixed data * Wait: provide a delay for a certain amount of time or until a specified date / time * Parallel: begin parallel branches of execution * Map: dynamically iterate steps