AWS SOA-C02 Drill: EC2 Right-Sizing - Balancing Cost and Performance

Jeff’s Note

#

Unlike generic exam dumps, ADH analyzes this scenario through the lens of a Real-World Site Reliability Engineer (SRE).

For SOA-C02 candidates, the confusion often lies in how to interpret CPU and memory utilization metrics to right-size EC2 instances effectively. In production, this is about knowing exactly which instance family and sizing adjustments reduce cost yet maintain performance. Let’s drill down.

The Certification Drill (Simulated Question)

#

Scenario

#

FinEdge Analytics runs a critical stateful workload continuously on an xlarge General Purpose On-Demand EC2 instance. CloudWatch monitoring shows the application uses approximately 80% of available memory and only 40% of CPU resources consistently. The SRE team must reduce instance costs without compromising workload performance.

The Requirement:

#

Identify the best EC2 instance type change that meets these performance patterns while lowering cost impact.

The Options:

#

• A) Switch to a large Compute Optimized On-Demand instance.
• B) Switch to a large Memory Optimized On-Demand instance.
• C) Switch to an xlarge General Purpose Spot instance.
• D) Switch to two large General Purpose On-Demand instances.

Google adsense

#

leave a comment:

Correct Answer

#

B) Switch to a large Memory Optimized On-Demand instance.

Quick Insight: The SOA-C02 Imperative

#

• Managing stateful instances means memory bottlenecks affect performance more critically than CPU under-utilization.
• Compute Optimized instances reduce CPU cost but won’t help memory constraints.
• Spot instances risk interruption which can impact availability for stateful workloads.
• Splitting the workload into two smaller instances doubles operational overhead and might increase cost.

Content Locked: The Expert Analysis

#

You’ve identified the answer. But do you know the implementation details that separate a Junior from a Senior?

The Expert’s Analysis

#

Correct Answer

#

Option B

The Winning Logic

#

The workload is memory bound — utilizing 80% of memory but only 40% CPU. This indicates the current instance’s CPU is under-utilized while memory capacity is near saturation.

• By switching to a large Memory Optimized instance, you match the high memory requirement while reducing CPU capacity and instance size from xlarge to large, cutting cost.
• Memory Optimized instances offer more RAM per vCPU compared to General Purpose.
• On-Demand ensures availability for your stateful workload without interruption.

The Trap (Distractor Analysis):

#

• Why not A (Compute Optimized)? This instance type offers more CPU but less memory relative to CPU, worsening the memory constraint.
• Why not C (Spot)? Spot instances can be interrupted; unsuitable for stateful critical workloads requiring constant uptime.
• Why not D (Two large General Purpose)? Running two instances increases complexity and could cost more; moreover, workload partitioning may be non-trivial if stateful.

The Technical Blueprint

#

# Example CLI command to check instance types with memory and CPU specs
aws ec2 describe-instance-types --filters Name=instance-type,Values=m5.large,m5.xlarge --query 'InstanceTypes[*].[InstanceType,MemoryInfo.SizeInMiB,VCpuInfo.DefaultVCpus]' --output table

The Comparative Analysis

#

Real-World Application (Practitioner Insight)

#

Exam Rule

#

“For the exam, always pick an instance type that aligns with the dominant resource utilization number.”

Real World

#

In practice, you might use Autoscaling with mixed instance types combined with Spot fleets for cost efficiency, but reliability is paramount for stateful applications so On-Demand memory-optimized is the safer bet.

(CTA) Stop Guessing, Start Mastering

#

Disclaimer

This is a study note based on simulated scenarios for the SOA-C02 exam.