The data engineer needs to orchestrate ETL jobs that include Spark jobs on Amazon EMR, API calls to Salesforce, and loading data into Redshift. They also need automatic failure handling and retries.Amazon Managed Workflows for Apache Airflow (Amazon MWAA)is the best solution for this requirement.

Option A: Amazon Managed Workflows for Apache Airflow (Amazon MWAA)Apache Airflow is designed for complex job orchestration, allowing users to define workflows (DAGs) in Python. MWAA manages Airflow and its integrations with other AWS services, including Amazon EMR, Redshift, and external APIs like Salesforce. It provides automatic retry handling, failure detection, and detailed monitoring, which fits the use case perfectly.

Option B (AWS Step Functions)can orchestrate tasks but doesn't natively support complex workflow definitions with Python like Airflow does.

Option C (AWS Glue)is more focused on ETL and doesn't handle the orchestration of external systems like Salesforce as well as Airflow.

Option D (Amazon EventBridge)is more suited for event-driven architectures rather than complex workflow orchestration.

A “No space left on device” error typically results fromdata skeworlarge shuffle stages. The best actions are:

B. Monitor using Spark UI and Glue metricsto find skewed partitions or executor issues.

D. Use --write-shuffle-files-to-s3to offload intermediate data to S3 instead of local disk, andapply saltingto reduce skew.

“You can reduce the impact of data skew and large shuffle operations by monitoring with Spark UI and enabling the --write-shuffle-files-to-s3 option. Salting can help rebalance the skewed keys.”

–Ace the AWS Certified Data Engineer - Associate Certification - version 2 - apple.pdf

Scaling out workers (A, C) is more costly and less efficient if the root cause (skew) is not fixed.

AWS Lambda is a serverless compute service that lets you run code without provisioning or managing servers. You can use Lambda to create functions that perform custom logic and integrate with other AWS services, such as API Gateway. Lambda automatically scales your application by running code in response to each trigger. You pay only for the compute time you consume1.

Amazon ECS is a fully managed container orchestration service that allows you to run and scale containerized applications on AWS. You can use ECS to deploy, manage, andscale Docker containers using either Amazon EC2 instances or AWS Fargate, a serverless compute engine for containers2.

Amazon EKS is a fully managed Kubernetes service that allows you to run Kubernetes clusters on AWS without needing to install, operate, or maintain your own Kubernetes control plane. You can use EKS to deploy, manage, and scale containerized applications using Kubernetes on AWS3.

The solution that meets the requirements with the least operational overhead is to create an AWS Lambda Python function with provisioned concurrency. This solution has the following advantages:

It does not require you to provision, manage, or scale any servers or clusters, as Lambda handles all the infrastructure for you. This reduces the operational complexity and cost of running your code.

It allows you to write your Python script as a Lambda function and integrate it with API Gateway using a simple configuration. API Gateway can invoke your Lambda function synchronously or asynchronously, and return the results to the frontend website.

It ensures that your Lambda function is ready to respond to API requests without any cold start delays, by using provisioned concurrency. Provisioned concurrency is a feature that keeps your function initialized and hyper-ready to respond in double-digit milliseconds. You can specify the number of concurrent executions that you want to provision for your function.

Option A is incorrect because it requires you to deploy a custom Python script on an Amazon ECS cluster. This solution has the following disadvantages:

It requires you to provision, manage, and scale your own ECS cluster, either using EC2 instances or Fargate. This increases the operational complexity and cost of running your code.

It requires you to package your Python script as a Docker container image and store it in a container registry, such as Amazon ECR or Docker Hub. This adds an extra step to your deployment process.

It requires you to configure your ECS cluster to integrate with API Gateway, either using an Application Load Balancer or a Network Load Balancer. This adds another layer of complexity to your architecture.

Option C is incorrect because it requires you to deploy a custom Python script that can integrate with API Gateway on Amazon EKS. This solution has the following disadvantages:

It requires you to provision, manage, and scale your own EKS cluster, either using EC2 instances or Fargate. This increases the operational complexity and cost of running your code.

It requires you to package your Python script as a Docker container image and store it in a container registry, such as Amazon ECR or Docker Hub. This adds an extra step to your deployment process.

It requires you to configure your EKS cluster to integrate with API Gateway, either using an Application Load Balancer, a Network Load Balancer, or a service of type LoadBalancer. This adds another layer of complexity to your architecture.

Option D is incorrect because it requires you to create an AWS Lambda function and ensure that the function is warm by scheduling an Amazon EventBridge rule to invoke the Lambda function every 5 minutes by using mock events. This solution has the following disadvantages:

It does not guarantee that your Lambda function will always be warm, as Lambda may scale down your function if it does not receive any requests for a long period of time. This may cause cold start delays when your function is invoked by API Gateway.

It incurs unnecessary costs, as you pay for the compute time of your Lambda function every time it is invoked by the EventBridge rule, even if it does not perform any useful work1.