Amazon AWS Certified Solutions Architect - Associate SAA-C03 Practice Test 51
Question 500
A network-intensive application has poor performance despite large instance CPU capacity. What EC2 capability should be checked?
- Use a spread placement group for supported small critical instance sets.
- Select an instance type with required enhanced networking and bandwidth characteristics.
- Balance cluster locality with capacity availability using supported placement strategies.
- Use a cluster placement group with compatible instances and AZ placement.
Explanation
Correct Answer
B:Network throughput ceilings differ among EC2 instance types and adapters. Benchmark instance network bandwidth and packet rates.
Incorrect Answers
A:Spread placement places instances on distinct underlying hardware where possible. Select an instance type with required enhanced networking and bandwidth characteristics. Benchmark instance network bandwidth and packet rates.
C:Cluster groups can constrain placement and instance capacity availability. Select an instance type with required enhanced networking and bandwidth characteristics. Benchmark instance network bandwidth and packet rates.
D:Cluster placement groups locate instances close for high-bandwidth low-latency networking. Select an instance type with required enhanced networking and bandwidth characteristics. Benchmark instance network bandwidth and packet rates.
Question 501
A tightly coupled HPC workload wants specialized low-latency communication fabric between instances. Which adapter is relevant?
- Use a partition placement group with supported node distribution.
- Use a spread placement group for supported small critical instance sets.
- Use Elastic Fabric Adapter on compatible EC2 instances and software stacks.
- Use a cluster placement group with compatible instances and AZ placement.
Explanation
Correct Answer
C:EFA provides low-latency networking support for supported HPC and ML workloads. Run an MPI bandwidth and latency benchmark.
Incorrect Answers
A:Partition placement reduces correlated rack failure for groups of instances. Use Elastic Fabric Adapter on compatible EC2 instances and software stacks. Run an MPI bandwidth and latency benchmark.
B:Spread placement places instances on distinct underlying hardware where possible. Use Elastic Fabric Adapter on compatible EC2 instances and software stacks. Run an MPI bandwidth and latency benchmark.
D:Cluster placement groups locate instances close for high-bandwidth low-latency networking. Use Elastic Fabric Adapter on compatible EC2 instances and software stacks. Run an MPI bandwidth and latency benchmark.
Question 502
A developer wants two EC2 instances placed in separate Availability Zones inside one cluster placement group. What topology constraint applies?
- Create a single cluster placement group and require its members to span both requested Availability Zones.
- Use a partition placement group with supported node distribution.
- Balance cluster locality with capacity availability using supported placement strategies.
- Use a suitable placement design because a cluster placement group is scoped to one AZ.
Explanation
Correct Answer
D:Cluster placement optimizes locality within an AZ rather than geographic availability. Inspect selected subnets before placing nodes.
Incorrect Answers
A:Cluster placement groups are scoped to one AZ, so the proposed two-AZ membership is not supported. A different placement strategy must be used for that cross-AZ requirement.
B:Partition placement reduces correlated rack failure for groups of instances. Use a suitable placement design because a cluster placement group is scoped to one AZ. Inspect selected subnets before placing nodes.
C:Cluster groups can constrain placement and instance capacity availability. Use a suitable placement design because a cluster placement group is scoped to one AZ. Inspect selected subnets before placing nodes.
Question 503
A compute fleet in one cluster placement group cannot scale rapidly due to insufficient contiguous capacity. What design tradeoff matters?
- Use a partition placement group with supported node distribution.
- Balance cluster locality with capacity availability using supported placement strategies.
- Use a suitable placement design because a cluster placement group is scoped to one AZ.
- Use edge/global routing such as CloudFront or Global Accelerator for client latency.
Explanation
Correct Answer
B:Cluster groups can constrain placement and instance capacity availability. Test scaling limits before finalizing the topology.
Incorrect Answers
A:Partition placement reduces correlated rack failure for groups of instances. Balance cluster locality with capacity availability using supported placement strategies. Test scaling limits before finalizing the topology.
C:Cluster placement optimizes locality within an AZ rather than geographic availability. Balance cluster locality with capacity availability using supported placement strategies. Test scaling limits before finalizing the topology.
D:EC2 placement groups optimize instance locality, not global client path selection. Balance cluster locality with capacity availability using supported placement strategies. Test scaling limits before finalizing the topology.
Question 504
A chat service needs cross-Region user latency optimization but its EC2 placement group is used as the proposed solution. What should change?
- Use a partition placement group with supported node distribution.
- Use edge/global routing such as CloudFront or Global Accelerator for client latency.
- Use a cluster placement group with compatible instances and AZ placement.
- Balance cluster locality with capacity availability using supported placement strategies.
Explanation
Correct Answer
B:EC2 placement groups optimize instance locality, not global client path selection. Measure end-user latency after adding the edge layer.
Incorrect Answers
A:Partition placement reduces correlated rack failure for groups of instances. Use edge/global routing such as CloudFront or Global Accelerator for client latency. Measure end-user latency after adding the edge layer.
C:Cluster placement groups locate instances close for high-bandwidth low-latency networking. Use edge/global routing such as CloudFront or Global Accelerator for client latency. Measure end-user latency after adding the edge layer.
D:Cluster groups can constrain placement and instance capacity availability. Use edge/global routing such as CloudFront or Global Accelerator for client latency. Measure end-user latency after adding the edge layer.
Question 505
A web fleet should maintain average CPU near 50% without manually calculating thresholds. Which policy fits?
- Review the Auto Scaling group's maximum size and service quotas for desired scale-out.
- Use scheduled scaling to increase fleet capacity before predictable demand.
- Configure EC2 Auto Scaling target tracking on average CPU utilization.
- Scale an Auto Scaling group on backlog per instance using suitable custom metrics.
Explanation
Correct Answer
C:Target tracking adjusts desired capacity toward a stated utilization objective. Apply load and monitor group activity.
Incorrect Answers
A:Scaling policies cannot exceed configured group capacity or regional quota limits. Configure EC2 Auto Scaling target tracking on average CPU utilization. Apply load and monitor group activity.
B:Scheduled capacity reduces cold-start delay for known traffic spikes. Configure EC2 Auto Scaling target tracking on average CPU utilization. Apply load and monitor group activity.
D:Backlog per worker relates capacity directly to outstanding jobs. Configure EC2 Auto Scaling target tracking on average CPU utilization. Apply load and monitor group activity.
Question 506
A queue-driven worker fleet should scale with messages per instance, not just average CPU. What metric helps?
- Tune metric targets, cooldown/warmup behavior and scale-in policy with real demand traces.
- Configure EC2 Auto Scaling target tracking on average CPU utilization.
- Review the Auto Scaling group's maximum size and service quotas for desired scale-out.
- Scale an Auto Scaling group on backlog per instance using suitable custom metrics.
Explanation
Correct Answer
D:Backlog per worker relates capacity directly to outstanding jobs. Increase queue depth and inspect scale-out timing.
Incorrect Answers
A:Inappropriate hysteresis and instance warmup can cause unstable scaling oscillations. Scale an Auto Scaling group on backlog per instance using suitable custom metrics. Increase queue depth and inspect scale-out timing.
B:Target tracking adjusts desired capacity toward a stated utilization objective. Scale an Auto Scaling group on backlog per instance using suitable custom metrics. Increase queue depth and inspect scale-out timing.
C:Scaling policies cannot exceed configured group capacity or regional quota limits. Scale an Auto Scaling group on backlog per instance using suitable custom metrics. Increase queue depth and inspect scale-out timing.
Question 507
A business receives regular large daily uploads at 7 AM. Which proactive scaling policy helps?
- Evaluate EC2 Auto Scaling predictive scaling for eligible historical load patterns.
- Configure EC2 Auto Scaling target tracking on average CPU utilization.
- Use scheduled scaling to increase fleet capacity before predictable demand.
- Scale an Auto Scaling group on backlog per instance using suitable custom metrics.
Explanation
Correct Answer
C:Scheduled capacity reduces cold-start delay for known traffic spikes. Compare request wait times before and after the schedule.
Incorrect Answers
A:Predictive scaling forecasts demand from past signals and can provision ahead of peaks. Use scheduled scaling to increase fleet capacity before predictable demand. Compare request wait times before and after the schedule.
B:Target tracking adjusts desired capacity toward a stated utilization objective. Use scheduled scaling to increase fleet capacity before predictable demand. Compare request wait times before and after the schedule.
D:Backlog per worker relates capacity directly to outstanding jobs. Use scheduled scaling to increase fleet capacity before predictable demand. Compare request wait times before and after the schedule.
Question 508
A retailer has strong recurring seasonal patterns and wants AWS to forecast needed EC2 capacity. What mechanism fits?
- Use scheduled scaling to increase fleet capacity before predictable demand.
- Configure EC2 Auto Scaling target tracking on average CPU utilization.
- Tune metric targets, cooldown/warmup behavior and scale-in policy with real demand traces.
- Evaluate EC2 Auto Scaling predictive scaling for eligible historical load patterns.
Explanation
Correct Answer
D:Predictive scaling forecasts demand from past signals and can provision ahead of peaks. Compare forecasts against actual utilization.
Incorrect Answers
A:Scheduled capacity reduces cold-start delay for known traffic spikes. Evaluate EC2 Auto Scaling predictive scaling for eligible historical load patterns. Compare forecasts against actual utilization.
B:Target tracking adjusts desired capacity toward a stated utilization objective. Evaluate EC2 Auto Scaling predictive scaling for eligible historical load patterns. Compare forecasts against actual utilization.
C:Inappropriate hysteresis and instance warmup can cause unstable scaling oscillations. Evaluate EC2 Auto Scaling predictive scaling for eligible historical load patterns. Compare forecasts against actual utilization.
Question 509
An Auto Scaling group is configured to scale out too aggressively and oscillates all day. What setting should be assessed?
- Scale an Auto Scaling group on backlog per instance using suitable custom metrics.
- Evaluate EC2 Auto Scaling predictive scaling for eligible historical load patterns.
- Tune metric targets, cooldown/warmup behavior and scale-in policy with real demand traces.
- Use scheduled scaling to increase fleet capacity before predictable demand.
Explanation
Correct Answer
C:Inappropriate hysteresis and instance warmup can cause unstable scaling oscillations. Inspect repeated scaling events after tuning.
Incorrect Answers
A:Backlog per worker relates capacity directly to outstanding jobs. Tune metric targets, cooldown/warmup behavior and scale-in policy with real demand traces. Inspect repeated scaling events after tuning.
B:Predictive scaling forecasts demand from past signals and can provision ahead of peaks. Tune metric targets, cooldown/warmup behavior and scale-in policy with real demand traces. Inspect repeated scaling events after tuning.
D:Scheduled capacity reduces cold-start delay for known traffic spikes. Tune metric targets, cooldown/warmup behavior and scale-in policy with real demand traces. Inspect repeated scaling events after tuning.
Question 510
A fleet's CPU is low but p95 response time is high due to I/O saturation. What should drive scaling decisions?
- Evaluate EC2 Auto Scaling predictive scaling for eligible historical load patterns.
- Configure EC2 Auto Scaling target tracking on average CPU utilization.
- Use meaningful application latency or throughput metrics instead of CPU alone.
- Scale an Auto Scaling group on backlog per instance using suitable custom metrics.
Explanation
Correct Answer
C:Scaling signals must reflect the actual bottleneck and user-facing objectives. Compare p95 latency before and after scaling.
Incorrect Answers
A:Predictive scaling forecasts demand from past signals and can provision ahead of peaks. Use meaningful application latency or throughput metrics instead of CPU alone. Compare p95 latency before and after scaling.
B:Target tracking adjusts desired capacity toward a stated utilization objective. Use meaningful application latency or throughput metrics instead of CPU alone. Compare p95 latency before and after scaling.
D:Backlog per worker relates capacity directly to outstanding jobs. Use meaningful application latency or throughput metrics instead of CPU alone. Compare p95 latency before and after scaling.