AWS Architecture Training

Agenda and Plan
Day 1:
Cloud and AWS : Overview
AWS Core Services.
Best Practices.
High Availability.

Day 2:
Scaling
Automation
Loose coupling
Storage.

Day 3:
AWS Architecture Framework
Troubleshooting
Large Scaling Platform management.

Student Guide Slide: Download from (Lumeris shared network folder)
P:\amahapatra\AWS Training guide v5.3
All the notes below are covered in details in this PDF

Cloud and AWS : Overview

• Why AWS. How Big. High level overview of services.
• Managed Service Vs Unmanaged service: Advantage and Disadvantages of these services on AWS.

AWS Core Services:

VPC:
. Relationship to Region and Availability Zones.

EC2:
. Which one to choose and why.
. Security Group
. Instance family groups, Optimised insances (network/EBS/etc)
. Feature of each family. (C5/ F1/ T2/etc)
. Pricing families and their features (On demand, Spot, Reserved and Dedicated Instances)
. Spot Instance and Hibernation agent- How to manage spot instance and back it up with EBS volumes .(find documents online)

S3:
. General feature and availability
. S3 notification and where to use them (Lambda, SQS)
. S3 Analytics for access patterns (read online)
. S3 Standard access, S3 Infrequent access
. Move objects based on age of storage through policies on bucket.

EBS:
. How to use with EC2 as a persistent boot volume.
. Snapshoting( point in time backup to S3) and its uses. (Unmanaged feature)
. Options of back-end hardware to choose (SSD/ HDD) and their use cases.

EFS:
. How it is different from EBS (access to multiple EC2)
. How to start/stop EFS connection to EC2

Glacier:
. Why differnet from s3
. Features

RDS:
. Managed service
. Supported Engines(Oracle , MySQL, etc)
. Scalability, Automated redundancy,
. When to / not to use RDS

DynamoDB:
. Managed service.
. On-demand backups
. Latency feature.
. Global Tables and their features (availability across regions)
. Read/Write capacity unit
. Auto Scaling throughput .

AWS IAM:
. IAM entities (Account, Users, Groups,Roles )and policies
. Roles vs policies
. Pass role and when to use it
. Policies (allow/deny)
. Federated user — Attached to Role — Roles gets policies
. Security handshakes and STS(security token service)
. Cloudtrail and its use cases.

AWS Lab 1:(30 Mins)

Deploy a php app on EC2, fetching content from s3, DynamoDB.
Overview of steps:
. Create IAM role and give it access to s3 and dynamoDB
. Create a s3 bucket
. Create a DynamoDB table
. Create a VPC, Subnet, Internet Gteway.
. Create Public route table.
. Launch an EC2 instance and in the start-up script, copy files to S3 and DynamoDB table.
. Examine the public IP of instance to reflect all resources of S3 and DynamoDb table.

AWS Best Practices

. Choosing Region (Price and Proximity)
. Choose AZ (how many should we have)
. VPC (how many and why)
. VPC design pattern (multi VPC design/ multi Account design)
. NAT Instance/ NAT Gateway / Internet Gateway
. Virtual Private Gateway
. VPC Peering
. Security groups (public and private)
. Elastic Network Interface. ( Elastic IP )
. Network Access Control List
. AWS Direct Connect

AWS Training day-2

AWS High Availability

. Services which are inherently available (s3, cloudformation) and not inherently avaialble(EC2,etc)
ELB:
.Classic LB vs Application LB vs Network LB
. Cross zone load balancing
. Connection draining
. Internal vs External LB

Elastic IP:
. May also get ENI for application need

Route 53:
. Latency based routing
. Geolocation routing.
. Failure over based routing

AWS Direct Connect:
- Use cases

AWS LAB 2: (45–60 mins):

Create AMI and launch a new instance.
Expand VPC to include more AZ.
Launch new insances into new AZ
Create VPC Subnets, NAT tables and route tables
Create Load Balancer and Auto scaling groups .

AWS Automation:

AWS Cloudwatch:
. Predefined hypervisor metrics
. Custom metrics
. Cloudwatch alarm + Autoscaling

AWS AutoScaling:
. through Cloudwatch
. through Scheduling
. automated autoscaling
. valid scenarios and autoscaling thrash

EC2 Auto Recovery:
. triggering through cloudwatch

Scaling RDS:
. scaling instance sizes(vertical)
. horizontal scaling through replica

AutoScaling DynamoDB:
. scaling through cloudwatch

AWS Lambda:
. scheduled
. parameters
. scenarios

AWS LAB 3: (30 mins):

Create a Lambda function. Create a SNS topic subscribed to an Autoscaling group. Trigger the Autoscaling. SNS should capture the new instance info and trigger a lambda which then snapshots the volume attached to the new instance.

Automating Infrastructure

Cloud formation
. resource grouping/sharing (security. )
. careful about sharing templates
. features: parameter, mapping,resources, etc
. custom resource
. CHAINING STACK (trigger one stack from another)

Beanstalk:
. feature and generated cloudformation
. blue green deployments

AWS OpsWork:
. Recipes / Cook book

EC2 Run command:

CLOUDFORMER:
. Reverse Engineering a cloud formation from resources.

Loose Coupling:

. Lambda serverless provisioning
. SOA to microservices
. going stateless
. Communicating with SQS for async processing.
SQS:
. Features(buffer batch operation, fan out, auto scaling the servers watching the queue)
.Short polling/ long polling
. visibility timeout
. security and encryption
SNS:
. non persistent queue.
. protocols
DynamoDB:
API Gateway:

Lambda:
. AWS Lambda step function : chaining multiple lambda of higher processing time.

AWS Training day-3

AWS Architecture Framework :

5 Sections:

1. Operational Excellence
   AWS CICD framework
   Cloud Formation
   Cloudwatch
   Cloudtrail
2. Security
   AWS Gaurd duty
   AWS Shield
   AWS Inspectors
   AWS Key Management Service
   AWS CloudHSM
   AWS Iam
   AWS Directory service….
3. Reliability
   SnowBall — usps data to aws
   pilot light
   low capacity stand by
   multi site active active
   disaster recovery plans and runs
4. Performance Efficiency
   Instance size
   Storage selection and recovery
5. Cost Optimization
   . Spot instance
   . Trusted Advisor
   .

AWS Lab 5:

Make a test cross region deployment and automatic fail over through route 53

Troubleshooting Scenarios:

1. Instance timeouts
2. poor network performance
3. I/O on EBS is low
4. CPU load on RDS too high
5. Access Denied

Sample Architectures :

Scenarios in diagram.

AWS Web-Scale Storage:

. Storing static contents on S3 and dynamic generation on EC2.
. S3 versioning.
. S3 Key design
. cloudfront and RTMP distribution
. NoSql vs Sql scaling
. dynamo dB features and best practices.
. RDS Security groups
. Aurora and its features

Lab — 4: