FSx for Lustre

Question 1

Which AWS service designed for High-Performance Computing (HPC), offers a POSIX-compliant, managed file system with read-after-write consistency and file locking support?

Answer 1

FSx for Lustre

Question 2

What are the performance capabilities of AWS FSx for Lustre?

Answer 2

• Data transfer speeds, scales to hundreds of GB/s in throughput
• Latency less than 1 ms

Question 3

What types of systems are compatible with AWS FSx for Lustre?

Answer 3

Linux-based systems

Question 4

What permissions model does AWS FSx for Lustre use?

Answer 4

POSIX-style permissions model

Question 5

How can AWS FSx for Lustre be accessed?

Answer 5

As a Private service it can be accessible:
- Within a VPC
- Through VPN connection
- Through AWS Direct Connect

Question 6

What are the primary use cases for AWS FSx for Lustre?

Answer 6

• High-Performance Computing (HPC)
• Machine Learning workloads
• Big Data analytics
• Video processing
• Financial modeling simulations

Question 7

What are the deployment options for AWS FSx for Lustre?

Answer 7

• Scratch: for short-term storage and high-performance workloads
• Persistent: for long-term storage

Question 8

Which AWS FSx for Lustre deployment type is optimized for short-term storage, prioritizing performance over durability, and may result in data loss due to the lack of replication, high availability, and resiliency features?

Answer 8

Scratch, where increased servers and disks elevate the risk of hardware failure leading to data loss

Question 9

Which AWS FSx for Lustre deployment type offers long-term storage, supports replication within a single AZ, protects against hardware failure, and automatically heals when failures occur?

Answer 9

Persistent, ensuring high availability within a single AZ

Question 10

What backup options are available for AWS FSx for Lustre?

Answer 10

Both deployment types support data backup to S3 using:
- Manual backups
- Automatic backups (retention period: 0–35 days, 0 - means automatic backup is disabled)

Question 11

When AWS FSx for Lustre is deployed, which AWS service can optionally be associated as a repository to store object data?

Answer 11

Amazon S3 can be associated as a data repository:
- Data is stored in S3 but visible within the file system
- Data is lazy-loaded into the file system from S3 upon access

Question 12

How can changes made to individual files within an AWS FSx for Lustre file system be synced back to the associated S3 data repository?

Answer 12

By using the command:
sudo lfs hsm_archive path/to/export/file

Question 13

Into which two components does AWS FSx for Lustre divide data when stored on disk?

Answer 13

• Metadata Storage Targets (MSTs): store file metadata, including filenames, timestamps, and permissions
• Object Storage Targets (OSTs): store the actual file data, with each object being 1.17 TiB in size

Question 14

What determines the baseline performance of AWS FSx for Lustre file systems?

Answer 14

Baseline performance is tied to the size of the file system, starting at 1.2 TiB and increasing in increments of 2.4 TiB

• Scratch Deployment: base performance of 200 MB/s per TiB of storage
• Persistent Deployment: options of 50 MB/s, 100 MB/s, or 200 MB/s per TiB of storage
• Both Deployments: support bursting up to 1,300 MB/s per TiB using a credit system

Question 15

Which AWS managed file system service is ideal for HPC, ML, and Big Data workloads?

Answer 15

FSx for Lustre