DOP-C02 Exam Dumps : AWS Certified DevOps Engineer - Professional

The question emphasizes “MOST securely” and involves a multi-account backup strategy with a centralized backup account. AWS Backup best practice for high-security environments is to use customer managed KMS keys for encrypting backup vaults rather than AWS managed keys. Customer managed keys provide tighter control over key lifecycle, key policies, and cross-account access.

Option A creates encrypted backup vaults in both the source and centralized accounts, each protected by a customer managed KMS key in that account. AWS Backup is then configured to create full EC2 backups (AMIs) and copy them to the centralized backup vault. This design ensures:

Encryption at rest in both accounts.

Independent key control and policies per account.

Secure cross-account backup copy behavior governed by explicit KMS key grants and vault access policies.

Option B incorrectly uses the source account’s KMS key to encrypt vaults in both accounts, which is not the recommended pattern; each account should manage its own CMK. Options C and D rely partially or fully on AWS managed keys, which are less appropriate when the requirement explicitly stresses maximum security and control.

Therefore, Option A best satisfies both the RPO/RTO goals (daily AMI backups with rapid restore) and the strongest encryption posture.

SCPs (Service Control Policies) are the best way to restrict permissions at the organizational level, which in this case would be used to restrict modifications to the IAM role used by the auditing application, while still allowing trusted administrators to make changes to it. Options C and D are not as effective because IAM permission boundaries are applied to IAM entities (users, groups, and roles), not the account itself, and must be applied to all IAM entities in the account.

Verified Answer: A, C, and E.

Short Explanation: To improve the performance of the application, the DevOps engineer should use Amazon RDS Proxy, implement the database connection opening outside the Lambda event handler code, and connect to the proxy endpoint from the Lambda function.

Amazon RDS Proxy is a fully managed, highly available database proxy for Amazon Relational Database Service (RDS) that makes applications more scalable, more resilient to database failures, and more secure1. By using Amazon RDS Proxy, the DevOps engineer can reduce the overhead of opening and closing connections to the database, which can improve latency and throughput2.

The DevOps engineer should connect the proxy to the Aurora cluster reader endpoint, which allows read-only connections to one of the Aurora Replicas in the DB cluster3. This can help balance the load across multiple read replicas and improve performance for read-intensive workloads4.

The DevOps engineer should implement the database connection opening outside the Lambda event handler code, which means using a global variable to store the database connection object5. This can enable connection reuse across multiple invocations of the Lambda function, which can reduce latency and improve performance.

The DevOps engineer should connect to the proxy endpoint from the Lambda function, which is a unique URL that represents the proxy. This can allow the Lambda function to access the database through the proxy, which can provide benefits such as connection pooling, load balancing, failover handling, and enhanced security.

The other options are incorrect because:

Implementing database connection pooling inside the Lambda code is unnecessary and redundant when using Amazon RDS Proxy, which already provides connection pooling as a service.

Implementing the database connection opening and closing inside the Lambda event handler code is inefficient and costly, as it can increase latency and consume more resources for each invocation of the Lambda function.

Connecting to the Aurora cluster endpoint from the Lambda function is not optimal for read-only queries, as it can direct traffic to either the primary instance or one of the Aurora Replicas in the DB cluster. This can result in inconsistent performance and potential conflicts with write operations on the primary instance.