Module 2: Compute in the Cloud

Compute in the Cloud

Create virutal machines to run tasks online

Compute

Processing power needed to run apps, manage data, and perform calculations

In the cloud, its available on demand and can be accessed remotely without physical hardware

Amazon EC2

instances with flexibility, cost effectivity, and fast speed

On demand compute capacity and costs are based on active usage

Configure security, network and storage

Challenges of On-Premise Instances

must purchase hardware, wait for delivery, and handle installation and configuration

Benefits of Cloud Instances

quick launch, scale, and stop set up

no delays or up front costs

EC2 Process

1. launch

2. connect

3. use instance

Launching an EC2

select AMI and choose instance type

AMI

AWS Machine Image, preconfigures the operating system, permissions, and additional software

Can use 1 AMI to launch multiple instances

Instance Type

determines the hardware, for CPU, memory and network

EC2 Connects what for Linux

SSH

EC2 Connects what for Windows

RDP (Remote Desktop Protocol)

EC2 Connects what for Simplified

AWS Systems Manager

What is an EC2 Instance

Virtual machine that can run commands, perform tasks, add storage, organize files

VM

Virtual Machine

EC2 Multi-Tenancy

shares physical host with other instances; isolated but can share resources

EC2 Hypervisor

In charge of resource sharing and isolation, and it runs on a host machine

EC2 Vertical Scaling/ Scale Up

Adjust instance size given usage by adding more power to the machines

EC2 Horizontal Scaling/ Scale Out

Adjust instance size given usage by adding more machines/ resources

EC2 Instance Types

1. General Purpose

2. Compute Optimized

3. Memory Optimized

4. Accelerated Computing

5. Storage Optimized

General Purpose EC2

balanced mix of compute, memory, and network

ideal if you don’t know how workloads will perform / diverse workloads

ie: web services, code repositories

Compute Optimized EC2

ideal for compute intensive tasks

ie: gaming servers, high performance computing, ML, scientific models

Memory Optimized EC2

ideal for memory-intensive tasks that need fast performance

ie: datasets, databases, data analytics

Accelerated Computing EC2

Uses hardware accelerators like GPUs to handle tasks (fast computing)

ie: floating point calculations, graphics, machine learning

Storage Optimized EC2

ideal for workloads that require high performance for locally stored data

ie: large databases, data warehouses, I/O intensive projects

API Calls

Application Programming Interfaces: Have predetermined ways they are used to interact with services

AWS Management Console

Manage resources visually, for beginners to build knowledge

human controlled dashboard, view AWS Bills, set up testing env, monitor resources and tasks

AWS CLI to invoke AWS API

Command Line Interface, uses a terminal and text based commands to invoke the AWS API

AWS SDK to invoke AWS API

Software Developer Kit, using programming languages and calls API directly in code

AWS CloudShell

Cloud based terminal with AWS CLI

EC2 Customer Responsibility

all security and management tasks for everything (OS, config, network, data)

EC2 AWS Responsibility

provide the software and hardware for EC2

Are EC2s managed?

No, they are unmannaged, it is mainly the customers responsibility

Steps to Launch an EC2

1. launch instance

2. choose name

3. choose AMI

4. choose instance type

5. choose key pair

6. choose network settings

7. choose storage option

8. opt. (add scripts)

Key Pair for EC2

2 keys, a public key which is in the EC2 instance, and a private key only you have, used for logging into the instance

Custom AMI

create your own custom tailored AMI

Pre-Configured AMI

use a provided AWS AMI for common operating systems and software

Third Party AMI

buy a third party AMI from Amazon Marketplace that someone else created

AMI Repeatability

because configurations are constant and deployment is automated, you can have a consistent testing and development environment

this reduces errors and helps with scaling

EC2 On Demand Instance Pricing

pay what you use only, no up front pay or commitment

EC2 Reserved Instance Pricing

save up to 75% by committing to a 1 year or 3 year term

for predictable workloads with specific instances or regions

EC2 Spot Instance Pricing

bid on spare compute capacity up to 90% off

have to be flexible to interuptions if and when AWS reclaims it

EC2 Savings Plan Instance Pricing

commit to consistent usage (compute power / hour) of 1 or 3 years, save up to 72%

EC2 Dedicated Instance Pricing

instances running on hardware dedicated solely to you, but you cant choose the physical server or have the whole server to yourself

isolation from other AWS customers

EC2 Dedicated Host Pricing

reserve an entire physical server for your exclusive use, full control

ideal for strict licensing or security needs

Capacity Reserve Pricing

reserve a certain capacity, and pay all of the capacity whether used or not

Reserved Instance Flexibility Pricing

for steady workloads with predcitable use, up to 75% savings

Scalability

handle increased or decreased loads by changing number of resources, focuses on long term capacity

Elasticity

automatically scale resources up and down, real time, rapid, on demand

Dynamic Scaling

auto adjusts in real time

Predictive Scaling

preemptively predicts scaling and follows trends

Minimum Capacity

least # of EC2s needed to run, assures system doesnt accidentally scale below threshold

Maximum Capacity

most # of EC2s to fit in budget, assures system doesnt accidentally scale above budget

Desired Capacity

ideal # of EC2s for workload, aims to stay at desired capacity

if no desired capacity is set, desired capacity = minimum capacity

Load Balancer

directs requests to instances like a host to avoid idle/ overworked instances

ELB

Elastic Load Balancing, distributes network traffic for good scalability and doesnt add to hourly costs

uses single URL for each frontend instance and directs the traffic to the least busy backend instance

Round Robin Routing Method

distributes traffic evenly across all servers in a loop

Least Connections Routing Method

routes traffic to the server with the fewest active connections

IP Hash Routing Method

uses the clients ip to route the traffic to the same server everytime

Least Response Time Routing Method

routes traffic to the server with the least response time for faster routing

Tightly Coupled

Loosely Coupled

Monolithic Application

Microservices Architecture

Amazon SQS

Simple Queue Service, lets software send store and recieve messages in a queue so no instance has to wait if another is down

Amazon SNS

Simple Notification Service, mass sends notification to all instances, can be done instantly, or take time

Amazon EventBridge

Event Traffic Control, rule based distribution of tasks

ie if event type is _____, send to _____, _____, _____,