Скачать или смотреть Tough AWS Solution Architect Certification Professional Exam Question on EC2 & Cluster Placement

❓ Scenario Recap
You have 9 EC2 instances running inside a Cluster Placement Group (CPG) and want to add two more. The new instances fail to launch due to a capacity error.

✅ Correct Answer:
Halt and restart all the running instances in the Cluster Placement Group, then try adding the new instances again.

🔍 Why This Works
A Cluster Placement Group attempts to pack instances close together within a single Availability Zone to maximize low-latency networking and high throughput—ideal for HPC (High Performance Computing) or real-time workloads.

When you initially launch instances together, AWS tries to place them on the same hardware rack or within contiguous capacity.

Adding instances later risks hitting hardware fragmentation—i.e., not enough neighboring capacity remains to fit more instances into that tightly packed group.

By stopping all existing instances and then restarting them along with the new instances, AWS gets a second chance to reallocate all of them together, possibly onto a new rack, resolving the capacity constraint.

❌ Why Other Answers Are Incorrect
“Launch in a separate placement group...”
While technically possible, this defeats the purpose of a Cluster Placement Group, which is to keep all instances tightly collocated for performance. Using different groups splits them physically, reducing the throughput benefits.

“Verify that all instances are of identical types...”
Important best practice, but not the root cause in this case. You’re getting a capacity error, not an instance-type mismatch.

“Contact AWS to increase the placement group limit...”
There's no hard limit on the number of instances in a Cluster PG; the constraint is available hardware capacity, not a configurable quota. A support ticket won’t help here.

💡 Tips for the AWS Solutions Architect Professional Exam
1. Understand the Three Placement Group Types
Cluster: High performance, low latency. Same AZ.

Partition: Used for big data (e.g., Hadoop), each partition isolated.

Spread: For critical apps; distributes instances across hardware.

You’ll often be tested on their trade-offs.

2. Know When to Use Restart Strategy
When dealing with Cluster Placement Groups, adding instances later is fragile—always try to launch as many as possible at once.

If adding later fails due to capacity, stop all instances and restart to let AWS retry placement.

3. Design for Flexibility
If performance isn't mission-critical, use Auto Scaling Groups and Spread Placement for resilience.

Elastic Fabric Adapter (EFA) can also be used for HPC—but requires specific instance types and configurations.

4. Beware of Zonal Constraints
Cluster Placement Groups are AZ-bound. If you don’t plan ahead, you could block future scalability.

5. Instance Type Fragmentation Matters
Even if you’re using the same instance type, the availability at the moment you scale may vary.

Use Capacity Reservations if you know you’ll scale up later.

🔧 Real-World Tip
In real deployments where high-throughput is essential, provision all instances at once or use an Auto Scaling Group with Warm Pools to keep extra capacity ready without overpaying.

📘 Summary for Exam Prep
Cluster PG = performance over flexibility.

Capacity errors in PGs = relaunch all together.

Know how PG types differ, and their failure/scale-out behavior.

Focus on design trade-offs (performance vs. availability).