A company wants to run high performance computing (HPC) workloads on AWS for financial risk modeling. The workloads run on Linux, use hundreds of Amazon EC2 Spot Instances, are short-lived, and generate thousands of output files that must be stored persistently for analytics and future use. The company needs a cloud storage solution that allows copying on-premises data to long-term persistent storage accessible to all EC2 instances, and a high-performance file system integrated with persistent storage for reading and writing datasets and outputs. Which combination of AWS services meets these requirements?

Amazon FSx for Lustre integrated with Amazon S3 is the correct choice because FSx for Lustre is a high-performance file system optimized for HPC workloads, offering sub-millisecond latencies and high throughput. Its native integration with Amazon S3 allows it to act as a high-performance cache for S3 data, making it ideal for reading large datasets and writing thousands of output files persistently to S3. The short-lived Spot Instances can leverage FSx for Lustre for fast processing and S3 for long-term storage. FSx for Windows File Server is designed for Windows-based applications and SMB protocols, not Linux HPC. Amazon S3 Glacier is for archival storage, not high-performance file system access. An S3 bucket with a VPC endpoint integrated with EBS gp2 would not provide the shared, high-performance file system required for multiple EC2 instances in an HPC environment. EBS volumes are block storage attached to a single instance, not a shared file system.