Examkin

Question

Difficulty: HardOptimizing Compute and Storage Performance

An investment bank runs a daily risk analysis simulation on AWS. The application tier runs on a fleet of memory-optimized Amazon EC2 instances in an Auto Scaling group (ASG). Every day at 09:00 UTC, a massive batch of simulation queries is initiated, causing immediate CPU utilization spikes on both the application fleet and the database, which is hosted on Amazon RDS PostgreSQL. Additionally, during the simulation, the EC2 instances perform heavy write operations to write transient checkpoint files to attached EBS gp3 volumes, resulting in high write latency and disk queue length spikes. Which combination of architectural modifications will optimize the compute and storage performance of this platform?

  1. Configure a scheduled scaling action for the EC2 Auto Scaling group to scale out before 09:00 UTC. Deploy Amazon RDS PostgreSQL Read Replicas and route read queries to the read replica endpoints. Increase the provisioned IOPS and throughput of the attached gp3 volumes to accommodate checkpoint write requirements.Answer
  2. B
    Configure a scheduled scaling action for the EC2 Auto Scaling group to scale out before 09:00 UTC. Route the read queries to the standby instance of the RDS PostgreSQL Multi-AZ deployment to offload the primary instance. Increase the provisioned IOPS and throughput of the attached gp3 volumes to accommodate checkpoint write requirements.
  3. C
    Rely on dynamic target tracking scaling policies based on CPU utilization to scale out the EC2 Auto Scaling group at 09:00 UTC. Deploy an Application Load Balancer to route traffic to the EC2 instances, relying on the load balancer's automatic scaling to handle the instant load spike. Deploy Amazon RDS PostgreSQL Read Replicas and route read queries to the read replica endpoints.
  4. D
    Migrate the simulation runs to AWS Fargate to scale compute dynamically, and purchase EC2 Instance Savings Plans to optimize the compute cost of these Fargate tasks. Deploy Amazon RDS PostgreSQL Read Replicas and route read queries to the read replica endpoints.

Answer

Configure a scheduled scaling action for the EC2 Auto Scaling group to scale out before 09:00 UTC. Deploy Amazon RDS PostgreSQL Read Replicas and route read queries to the read replica endpoints. Increase the provisioned IOPS and throughput of the attached gp3 volumes to accommodate checkpoint write requirements.
The correct option addresses all three performance bottlenecks optimally. Scheduled scaling resolves the slow response of dynamic scaling before the daily 09:00 UTC simulation run. Deploying RDS Read Replicas offloads database CPU usage by distributing the read spike. Increasing provisioned IOPS and throughput on EBS gp3 volumes directly resolves storage latency bottlenecks without paying for unnecessary storage space.

Step-by-Step Solution

1
Address predictable compute scale-out latency.
Implement a scheduled scaling action for the Auto Scaling group set to trigger prior to 09:00 UTC.
Target tracking policies react only after metrics cross thresholds, leading to performance degradation during sudden spikes. Scheduled scaling ensures compute capacity is warm and ready before the workload begins.
2
Address database read bottlenecks.
Deploy one or more RDS PostgreSQL Read Replicas and modify the application to route read queries to their endpoints.
RDS PostgreSQL Multi-AZ standbys cannot handle read connections. Offloading query traffic to dedicated Read Replicas preserves primary instance CPU for write transactions.
3
Optimize transient storage performance.
Provision higher IOPS and throughput parameters directly on the gp3 volumes without expanding disk size.
EBS gp3 volumes allow independent adjustment of IOPS and throughput to meet temporary performance spikes, avoiding the need to over-provision storage capacity.

Key Concept

Optimizing compute and storage involves matching workloads to the correct provisioning mechanisms: scheduled scaling for predictable compute spikes, read replicas for database read distribution, and independent EBS throughput provisioning.
Rate this question

Topics

Optimizing Compute and Storage Performance