AUTO SCALING

Configure automatic scaling for AWS resources quickly that use Dynamic Scaling and Predictive Scaling

· optimize for availability or Cost or both

· scaling in means decreasing the number of instances while scaling out means increasing the number

· Useful for

o Cyclical traffic - Business hours / After hours

o On and Off traffic patterns - batch processing / testing

o Variable traffic patterns - Marketing campaigns

· region specific service

Features

· Launch or Terminate EC2 instances in an auto scaling group

· Adjust ECS up or down

· enable DynamoDB table or global secondary index increase/decrease read/write capacity without throttling

· dynamically adjust Aurora number of read replicas

· Dynamic scaling

· Predictive Scaling

· Auto scaling scans the environment and automatically discovers scalable cloud resources underlying the application so you don't have to manually identify them

EC2 Auto Scaling

· Groups

· Launch configurations

· Scaling options

o Scale to maintain current instance levels at all times

o Manual scale

o Scale based on schedule

o Scale based on Demand

lifecycle hook : to perform actions when scaling in/out (i.e copy EBS data to S3 before termination etc)

· Scaling policy types

o Target Tracking Scaling : use a metric that is monitored that triggers the dynamic scaling

o Step Scaling

o Scheduled scaling

o Scale cooldown period (in secs)

· Termination of instances

o

Application Auto Scaling

· Allows to auto scale for the following AWs resources

o ECS

o Spot Fleet requests

o EMR

o AppStream 2.0

o DynamoDB Tables and Global Indexes

o AUROA read replicas

o SageMaker endpoint variants

o custom resources by your own applications or services

· Features

o Target Tracking Scaling :

o Step Scaling

o Scheduled scaling

o Scale in cooldown timers (in secs)

o Scale Out cooldown period (in secs)

· Can attach one or more Classic ELBs to ASG , must be in same region

· ASG rebalances by scaling out to the AZ`s that have fewer instances first

Monitoring

· Health checks

· CloudWatch metrics

· CloudWatch events

· SNS notifications

Troubleshooting

· Can suspend resume auto scaling when troubleshooting an issue

Security

· IAM

Limits