SAP-C02 · Question #195
A greeting card company recently advertised that customers could send cards to their favorite celebrities through the company's platform. Since the advertisement was published, the platform has receiv
The correct answer is B. Use Amazon RDS Proxy. Reconfigure the database connections to use the proxy.. RDS Proxy is correct because the symptom - Aurora connection failures with low CPU - is a classic connection exhaustion problem. With 10,000 users/second each spawning a new process and holding an open MySQL connection for the session, the platform quickly exceeds Aurora's max_co
Question
A greeting card company recently advertised that customers could send cards to their favorite celebrities through the company's platform. Since the advertisement was published, the platform has received constant traffic from 10,000 unique users each second. The platform runs on m5.xlarge Amazon EC2 instances behind an Application Load Balancer (ALB). The instances run in an Auto Scaling group and use a custom AMI that is based on Amazon Linux. The platform uses a highly available Amazon Aurora MySQL DB cluster that uses primary and reader endpoints. The platform also uses an Amazon ElastiCache for Redis cluster that uses its cluster endpoint. The platform generates a new process for each customer and holds open database connections to MySQL for the duration of each customer's session. However, resource usage for the platform is low. Many customers are reporting errors when they connect to the platform. Logs show that connections to the Aurora database are failing. Amazon CloudWatch metrics show that the CPU load is low across the platform and that connections to the platform are successful through the ALB. Which solution will remediate the errors MOST cost-effectively?
Options
- ASet up an Amazon CloudFront distribution. Set the ALB as the origin. Move all customer traffic to
- BUse Amazon RDS Proxy. Reconfigure the database connections to use the proxy.
- CIncrease the number of reader nodes in the Aurora MySQL cluster.
- DIncrease the number of nodes in the ElastiCache for Redis cluster.
How the community answered
(45 responses)- A29% (13)
- B56% (25)
- C4% (2)
- D11% (5)
Explanation
RDS Proxy is correct because the symptom - Aurora connection failures with low CPU - is a classic connection exhaustion problem. With 10,000 users/second each spawning a new process and holding an open MySQL connection for the session, the platform quickly exceeds Aurora's max_connections limit. RDS Proxy sits between the application and Aurora, pooling and multiplexing connections so thousands of app-side connections share a much smaller set of actual database connections - solving the issue without adding expensive compute.
Option A (CloudFront) is wrong because CloudFront is a content delivery network for caching and edge distribution; it has no effect on database connection limits. Option C (more Aurora reader nodes) doesn't fix the root cause - each reader node has its own connection ceiling, and the problem is that connections aren't being released, not that read capacity is insufficient; it also adds ongoing cost. Option D (more ElastiCache nodes) is irrelevant because the errors are in Aurora, not Redis.
Memory tip: When you see low CPU + connection failures + long-lived per-user DB connections, think "Proxy pools" - RDS Proxy is the targeted, cost-effective fix for connection exhaustion because it multiplexes at the proxy layer rather than scaling the database itself.
Topics
Community Discussion
No community discussion yet for this question.