AWS Auto Scaling Group Continued
- Impact of parallel connections to a server. As users create parallel/concurrent connections (session) the load on cpu increases.
- In this lets try doing this,
- increase number of ec2 instances by 1 when average cpu utilization for last 5 minutes > 70
- decrease number of ec2 instances by 1 when average cpu utilization for last 5 minutes < 50
- minimum instances = 1, max instances = 5
- To create artificial stress on the ec2 instance we will be using stress tool.
- Steps followed:
- Wait for the auto scaling group to be created with no scaling policy. Now lets create a dynamic scaling policy
- Now Select Metric and Create Cloudwatch Alarm
- Now lets generate artifical load. Login into ec2 instance
stress --cpu 8 --io 4 --vm 2 --vm-bytes 128M --timeout 60m -v
- Now observe for the changes in desired capacity in the period of 5-10 mins
- Now stop the stress tool (ctrl + c) and wait for 5-10 mins the desired capacity will be back to 1
