on-demand and autoscaling

[mddreeves]

In this question below, how would running 5 instances during normal operations be viable if 25 instances running at operations operate at 20% CPU average?

A startup is running a data processing application on AWS. The application is hosted on 25 Amazon EC2 On-Demand Instances, distributed across three Availability Zones, and registered with a target group for a Network Load Balancer (NLB).

Reports indicate a sharp decline in performance during high-demand periods, with CPU utilization spiking to 90%-100%. During normal operations, utilization averages only 20%. The application is stateless and needs to ensure consistent response times even during peak traffic.

The company wants to improve application performance while optimizing costs over the next three years.
Which solution is the most cost-effective for addressing this issue? (Answer is A)

A. Configure an Auto Scaling group and attach it to the NLB. Set the minimum capacity to 5 instances and the maximum capacity to 30. Purchase Reserved Instances for 5 instances.

B. Replace the On-Demand Instances with a Spot Fleet request type of request and with EBS-optimized enabled. Set the TotalTargetCapacity parameter to 25 and DefaultTargetCapacityType parameter to Spot. Attach the Spot Fleet to the NLB.

C. Replace the On-Demand Instances with a Spot Fleet request type of request. Configure the InstanceInterruptionBehavior to stop and TotalTargetCapacity parameter to 30. Attach the Spot Fleet to the NLB.

D. Configure an ASG (Auto Scaling Group) and attach it to the Network Load Balancer. Set the capacity to a minimum of 15 instances and a maximum of 25. Purchase Reserved Instances for 15 instances.

[JR-TutorialsDojo]

Hello mddreeves,

Thanks for reaching out.

Based on the given scenario, CPU usage during normal operations remains low at approximately 20 percent across 25 instances. This indicates that the application is not making full use of the available computing resources. With Auto Scaling in place, the system adjusts to workload demands in real time. The five Reserved Instances can efficiently support the baseline demand at a lower cost, while the system can scale out to up to 30 instances as needed during peak periods. This means during normal periods, resources aren’t wasted, and during heavy traffic, performance remains stable.

I hope this helps! Let us know if you need further assistance.

Regards,
JR @ Tutorials Dojo

[mddreeves]

I get the concept, for sure, and understand the need to ASG up to 30 instances for peak load. I guess what I am saying is that if 25 instances operate at 20% CPU under normal conditions, how would 5 instances be able to facilitate normal operations?

[JR-TutorialsDojo]

Hello mddreeves,

Thanks for the reply. Could you please elaborate a bit more on your question so I can make sure I’m addressing it accurately?

Regards,
JR @ Tutorials Dojo

[mddreeves]

Thanks, but I get it. Basically the 5 instances will operate at a much higher CPU load at normal operations than the 25 instances did, but that is satisfactory

[JR-TutorialsDojo]

Thanks, mddreeves!

Feel free to reach out if you have any more questions or need further assistance.