Which solution will meet the performance requirements?
Create an Amazon FSx for NetApp ONTAP file system. Sat each volume’ tiering policy to ALL. Import the raw data into the file system. Mount the fila system on the EC2 instances.
Create an Amazon S3 bucket to store the raw data. Create an Amazon FSx for Lustre file system that uses persistent SSD storage. Select the option to import data from and export data to Amazon S3. Mount the file system on the EC2 instances.
Create an Amazon S3 bucket to store the raw data. Create an Amazon FSx for Lustre file system that uses persistent HDD storage. Select the option to import data from and export data to Amazon S3. Mount the file system on the EC2 instances.
Create an Amazon FSx for NetApp ONTAP file system. Set each volume’s tiering policy to NONE. Import the raw data into the file system. Mount the file system on the EC2 instances.
Explanations:
While Amazon FSx for NetApp ONTAP provides good performance, setting the tiering policy to ALL may lead to some data being stored in lower-speed tiers, which could affect latency and throughput requirements. Sub-millisecond latencies and 6 GBps throughput are better supported by directly managing data on SSDs.
Amazon FSx for Lustre using persistent SSD storage can provide the required sub-millisecond latencies and a throughput of 6 GBps. The integration with Amazon S3 allows for efficient data import/export, and mounting the file system on EC2 instances will facilitate high-performance processing.
Although Amazon FSx for Lustre supports integration with S3, using persistent HDD storage does not meet the latency requirements. HDDs typically do not provide the sub-millisecond latencies needed for this use case.
Amazon FSx for NetApp ONTAP can provide good performance, but setting the tiering policy to NONE means that all data must reside on higher-speed storage, which could potentially impact performance if data management is not optimized for throughput. It is not the best solution compared to the SSD-based options.
As far as I’m aware, the answer is:
Create an Amazon S3 bucket to store the raw data. Create an Amazon FSx for Lustre file system that uses persistent SSD storage. Select the option to import data from and export data to Amazon S3. Mount the file system on the EC2 instances.