Examkin

Soru

Zorluk: OrtaPerformance and Scalability Optimization

A financial institution is launching a new mobile banking application. During a scheduled national marketing campaign, the application expects a sudden, massive spike in traffic, transitioning from a baseline of 3,0003,000 requests per second (RPS) to a peak of 120,000120,000 RPS within a few minutes. The application's backend runs on Amazon ECS tasks using AWS Fargate, and uses an Amazon Aurora PostgreSQL database for transaction history. Which two actions should a solutions architect recommend to ensure the platform scales and maintains performance during the launch? (Select TWO.)

  1. Contact AWS Support to pre-warm the Application Load Balancer to the expected traffic volume of 120,000120,000 RPS.Cevap
  2. Configure scheduled scaling policies for the Amazon ECS service to scale out Fargate tasks to the anticipated peak capacity before the campaign begins.Cevap
  3. C
    Rely on the Application Load Balancer's automatic scaling to absorb the initial spike, and configure ECS target tracking policies based on CPU utilization to handle task scaling.
  4. D
    Configure Aurora Auto Scaling to dynamically add reader instances during the traffic spike to handle the surge in write transactions.
  5. E
    Enable Aurora Multi-AZ standby replication and configure the application to load-balance write operations between the primary instance and the standby instance.

Cevap

Contact AWS Support to pre-warm the Application Load Balancer to the expected traffic volume, and configure scheduled scaling policies for the Amazon ECS service to scale out Fargate tasks to the anticipated peak capacity before the campaign begins.
For predictable flash traffic spikes of this magnitude, standard reactive scaling is insufficient. The Application Load Balancer must be pre-warmed by contacting AWS Support to prevent dropped connections during the rapid increase. Similarly, the Amazon ECS service must use scheduled scaling policies to pre-provision Fargate tasks ahead of time, ensuring sufficient compute capacity is online when the traffic arrives.

Adım Adım Çözüm

1
Analyze the traffic profile and determine the scaling requirements.
The traffic increases from 3,000 RPS to 120,000 RPS (a 40x increase) in a few minutes, which constitutes flash traffic.
Standard reactive scaling mechanisms for ALBs and ECS tasks are too slow to cope with this rate of increase.
2
Address the entry point scaling (Load Balancing).
Requesting ELB pre-warming from AWS Support ensures the ALB is pre-provisioned with the capacity to handle 120,000 RPS.
Without pre-warming, the ALB will drop connections while attempting to scale out reactively.
3
Address the compute layer scaling (ECS/Fargate).
Create scheduled scaling policies to scale the ECS service's task count to the anticipated peak level before the event starts.
Scheduled scaling ensures compute capacity is ready immediately when the campaign begins, avoiding cold-start latency and delayed reaction times of target tracking policies.

Anahtar Kavram

Handling sudden, predictable flash traffic spikes requires proactive pre-provisioning (ELB pre-warming and scheduled scaling for compute layers) rather than relying on reactive auto-scaling.
Tahmini Süre:2m 0s
Bu soruyu puanla

Konular

Performance and Scalability Optimization

Bu konudan daha fazla soru