QuestionQ34

VerifiedDesign High-Performing Architectures

A weather forecasting company must process hundreds of gigabytes of data with sub-millisecond latency. The company has a high performance computing (HPC) environment in its data center and wants to expand its forecasting capabilities.

A solutions architect needs to identify a highly available cloud storage solution that can support large amounts of sustained throughput. Files stored in the solution must be accessible to thousands of compute instances that will simultaneously access and process the entire dataset.

What should the solutions architect do to meet these requirements?

Explanation

Amazon FSx for Lustre persistent file systems are built for HPC and scale-out workloads that need consistent sub-millisecond latency, very high sustained throughput, and parallel access from large numbers of compute instances. Persistent FSx for Lustre is the highly available option because data is replicated and failed file servers are automatically replaced. Scratch FSx for Lustre is intended for temporary storage and is not replicated, while Amazon EFS Bursting Throughput is not appropriate for sustained high throughput because it relies on burst credits. Official AWS documentation describes persistent FSx for Lustre as ideal for longer-term, throughput-focused workloads and notes that the SSD storage class provides consistent sub-millisecond latency across the full dataset.

Learn more

Community Discussion

No comments yet. Be the first to start the discussion!