Scaling Compute: Elastic Load Balancing and Auto Scaling

Question 1

Design required for highly available and secure

website on EC2 with ALB, and DB on EC2

Answer 1

Launch ALB in public subnets, web servers in private subnets and DB layer in private subnets – all layers across AZs

Question 2

HealthyHostCount metrics for an ALB have dropped from 6 to 2. Need to determine the cause

Answer 2

The health checks on target EC2 instances are failing

Question 3

An instance attached to an ALB exceeded

the UnhealthyThresholdCount for consecutive health check failures. What will happen?

Answer 3

Health checks will continue and the ALB will take the instance out of service

Question 4

Requirement to track the source IP of clients and the instance that processes the request

Answer 4

Check the ALB access logs for this information

Question 5

Requirement to trigger an alarm when all instances are unhealthy

Answer 5

Use Amazon CloudWatch with the condition:

“AWS/ApplicationELB HealthyHostCount <= 0”

Question 6

Need to check why users cannot connect to web server public IP and port (behind ALB)

Answer 6

Check the VPC Flow Logs

Question 7

HTTPCode_ELB_5XX_Count Amazon CloudWatch metrics are noticed for an ALB

Answer 7

The target group may not contain any healthy

instances

Question 8

CloudWatch shows 4XX errors for app with ALB but the Instances have already been terminated and need to analyze the root cause

Answer 8

Use ELB access logs to retrieve info from S3 bucket to find the originators of the requests

Question 9

Need a load balancer where specific static public IP addresses can be whitelisted by clients

Answer 9

Use a Network Load Balancer (NLB)

Question 10

Poor performance has been experienced for an application running on Amazon EC2

Answer 10

Use EC2 Auto Scaling to dynamically scale

Question 11

503 and 504 errors experienced and instances have high CPU utilization

Answer 11

Use EC2 Auto Scaling to dynamically scale

Question 12

ASG does not launch instances during busy periods despite max capacity not being reached

Answer 12

Could be due to service limits (check Trusted Advisor) or check for RunInstances requests in CloudTrail in case they are failing

Question 13

Need to analyze instances before they are terminated

Answer 13

Use Auto Scaling lifecycle hooks to pause termination

Question 14

Auto Scaling scales based on queue depth but at beginning of day app slows down

Answer 14

Create a scheduled scaling policy

Question 15

Create highly available EC2 Auto Scaling group for a single instance app

Answer 15

Use at least 3 AZs, min size of 2, desired capacity of 2, and max of 2

Question 16

Elastic Beanstalk worker node reads messages from SQS queue. Auto Scaling scales instances. App slows down when number of messages in queue increases

Answer 16

Update ASG to scale on queue depth

Question 17

ALB is expecting a large spike in traffic and the application is memory heavy

Answer 17

Use the RequestCountPerTarget metric to control scaling

Question 18

New instances in an Auto Scaling group are not showing up in the aggregated metrics. Step scaling is used

Answer 18

Likely due to the warm-up period having not yet expired