A company has a data-ingestion workflow that uses an Amazon Simple Notification Service (Amazon SNS) topic to notify about new deliveries and an AWS Lambda function to process the data and record metadata. The workflow sometimes fails due to network connectivity issues; when it fails, the Lambda function does not ingest the data unless the company manually reruns the job. Which combination of actions should a solutions architect take so the Lambda function ingests all data reliably in the future? (Choose two.)

This solution addresses the reliability issue by decoupling the SNS topic from the Lambda function using an SQS queue. When the SNS topic publishes a message, the SQS queue receives and stores it. If the Lambda function experiences a transient failure (e.g., network connectivity), the message remains in the SQS queue, ensuring no data loss. The Lambda function is then modified to poll and process messages from the SQS queue. This provides a robust, asynchronous processing model with automatic retries and dead-letter queue capabilities, making the data ingestion reliable. Deploying Lambda in multiple AZs improves availability but doesn't guarantee data ingestion if the function fails to process a message. Increasing CPU/memory or provisioned throughput won't prevent data loss during transient failures.