Preparing metrics for Predictive scaling

Preparing data for Predictive scaling

Because Predictive scaling requires a large amount of data over more than 2 days to be able to make predictions for the following days, and here we do not have that data, we will need to prepare to simulate an environment

Preparation steps

First, we will create a new folder named metric-preparation and move it into this folder

mkdir metric-preparation && cd metric-preparation

Then download the script to prepare the data

curl -o prepare-metric-data.sh https://raw.githubusercontent.com/awslabs/ec2-spot-workshops/master/workshops/efficient-and-resilient-ec2-auto-scaling/prepare-metric-data.sh

role

After downloading, go inside to change the command section in the script

vim prepare-metric-data.sh

#!/bin/bash
set -e
file=$1
group=$2
echo "=== Format metric file ==="
echo $file
echo $group

l=$(jq length $file)
i=0
while [ $i -lt $l ]
do
time=$(date -d "$[5*$i] minutes ago")
cat $file | jq --argjson i $i --arg t "$time" '.[$i].Timestamp |= $t' > tmp.json && mv tmp.json $file
i=$[$i+1]
done

echo "replace autoscaling group name.."

sed -i $file -e "s/#ASGPLACEHOLDER#/$group/g"

echo "=== Complete ==="

role1

After editing, we now proceed to load the unprocessed data, first the metrics for the instances.

curl -o metric-instances.json https://raw.githubusercontent.com/awslabs/ec2-spot-workshops/master/workshops/efficient-and-resilient-ec2-auto-scaling/metric-instances.json

CPU Data

curl -o metric-cpu.json https://raw.githubusercontent.com/awslabs/ec2-spot-workshops/master/workshops/efficient-and-resilient-ec2-auto-scaling/metric-cpu.json

role1

Modify these two data types in turn, first for the CPU

bash prepare-metric-data.sh metric-cpu.json FCJ-Management-ASG && cat metric-cpu.json

createpolicy

Instances

bash prepare-metric-data.sh metric-instances.json FCJ-Management-ASG && cat metric-instances.json

createpolicy

💡 Note In the 2 commands above, the parameter FCJ-Management-ASG appears, which is the name of the Auto Scaling Group that we will create later, so later you will need to create an ASG with the same name. Otherwise, you should change to another name from now on.

Upload data to CloudWatch

In Amazon Linux 2023, and using the correct AMI, AWS CLI is already installed inside, at this time we just need to take it out to reconfigure the credentials. Remember that you must have an IAM User with enough permissions to upload data to CloudWatch or at least enough permissions to do this workshop.

Go to the IAM page, enter the IAM User information and enter the Access Key Id and Serect Access Key

aws configure

Configure

createpolicy

Then upload the 2 data files that we have prepared earlier to CloudWatch

aws cloudwatch put-metric-data --namespace 'FCJ Management Custom Metrics' --metric-data file://metric-cpu.json

aws cloudwatch put-metric-data --namespace 'FCJ Management Custom Metrics' --metric-data file://metric-instances.json

createpolicy

Check

Go to CloudWatch to check the results result

  • Find CloudWatch
  • Click to enter CloudWatch Console

createpolicy

In the Console interface of CloudWatch

  • Select All metrics
  • Select FCJ Management Custom Metrics

createpolicy

Select AutoScalingGroupName

createpolicy

Select 2 more parameters as shown below, wait a while to receive the results.

createpolicy