Question

Difficulty: HardModernizing Workloads with Serverless Architectures (Lambda and API Gateway)

A utility enterprise is modernizing its legacy smart meter telemetry ingestion backend to a serverless architecture on AWS. The solution must ingest bursty, real-time meter readings via an Amazon API Gateway endpoint and process them using an AWS Lambda function. The API Gateway endpoint must only be accessible privately from the enterprise's on-premises network via an AWS Direct Connect connection. The processed data is then stored in an Amazon RDS for PostgreSQL database in a private subnet. The database requires highly resilient connection management to prevent connection exhaustion during peak traffic times. Additionally, the ingestion workload must be configured so that it does not exhaust the regional execution pool and throttle critical downstream customer billing and payment Lambda functions. Which combination of actions should a solutions architect recommend to meet these requirements? (Select TWO.)

  1. Deploy a private API Gateway endpoint using an interface VPC endpoint, and configure an Amazon RDS Proxy between the telemetry processing Lambda function and the RDS for PostgreSQL database.Answer
  2. Configure reserved concurrency on the telemetry processing Lambda function to limit its maximum concurrent executions, preventing it from consuming the entire regional concurrency pool.Answer
  3. C
    Configure the telemetry processing Lambda function inside a private subnet, and route all outbound database traffic through a single NAT Gateway deployed in a single Availability Zone.
  4. D
    Store the database credentials in AWS Secrets Manager, encrypt them using the default AWS-managed KMS key, and configure the KMS key policy to grant decryption permissions to the Lambda function's IAM role.
  5. E
    Increase the default unreserved concurrency limit of the AWS account and configure provisioned concurrency on the customer billing and payment Lambda functions to absorb telemetry-induced bursts.

Answer

Deploy a private API Gateway endpoint using an interface VPC endpoint, configure an Amazon RDS Proxy between the telemetry processing Lambda function and the RDS database, and configure reserved concurrency on the telemetry processing Lambda function to limit its execution rate.
The correct answer combination involves deploying a private API Gateway with an interface VPC endpoint and RDS Proxy, and setting reserved concurrency on the telemetry function. The private API Gateway ensures that traffic from the on-premises network via AWS Direct Connect remains private. Amazon RDS Proxy pools connections to prevent connection exhaustion at the database layer. Configuring reserved concurrency on the telemetry Lambda function sets a concurrency ceiling, guaranteeing that the remaining regional concurrency is preserved for downstream payment and billing functions.

Step-by-Step Solution

1
Analyze private API connectivity requirements.
Identify that a private API Gateway combined with an interface VPC endpoint (AWS PrivateLink) allows the on-premises network to securely route traffic to API Gateway via AWS Direct Connect without traversing the public internet.
Meets the requirement for the API Gateway endpoint to be privately accessible from the corporate network.
2
Address database connection resiliency.
Determine that Amazon RDS Proxy should be placed between the Lambda function and the PostgreSQL database to pool and share database connections.
Prevents connection exhaustion on the RDS instance caused by highly concurrent Lambda scale-out events.
3
Design concurrency controls to protect neighboring workloads.
Apply reserved concurrency to the high-volume, bursty telemetry Lambda function.
Configuring reserved concurrency isolates the function's scale-out capacity, leaving the remaining regional account pool available for other critical functions like payment and billing.

Key Concept

Serverless architectural integration patterns using API Gateway private endpoints, RDS database connection pooling via RDS Proxy, and concurrency isolation via Lambda reserved concurrency.
Rate this question