- Storage Optimized: 80TB
- Compute Optimized: 42TB
- AWS DataSync
- 8 TB
- Use when transfer larger than 10 PB
- 100 PB max for single one
- The the computation at limited/no internet access place
- When send back data to AWS
- Snowball Edge -- storage clustering available
- Long term deployment
- You cannot directly import Snowball data into Glacier
- To S3 first, then use lifecycle policy
AWS Storage Gateway
- Bridge between on-premises data and cloud data in S3
- User cases: disaster recovery, backup & restore, tiered storage
- 3 types of Storage Gateway:
- File Gateway
- Volume Gateway
- Tape Gateway
- File access / NFS
- auth with Active Directory
- Storage backup, EBS Backup to S3
- On-premise data to the cloud => Storage Gateway
- NFS / Auth with Active Directory => File Gateway
- Volume / Block Storage / iSCSI => Volume Gateway
- Tape => Tape Gateway
- No on-premise virtualization => Hardware Appliance
- File gateway, NFS, SMB interface, backed by S3
- Volume Gateway, backed by S3, iSCSI interface
- Stored mode: mainly store the data on-premise
- Cached mode: only store cached file on-premise, mainly store in S3
- Store mode and Cache mode can change in time: for example, slowly migrate from on-premise to cloud by using Stored mode; when data is mainly on S3, then change to Cached mode, on-premise only keeps cached files.
- Tape gateway, iSCSI interface.
FSx for Windows
- EFS is a shared POSIX system for Linux system
- FSx for Windows is a fully managed Windows file system shared drive
- Can be configured to be Multi-AZ
- Data is backup daily to S3
- Can be accessed from your on-premise infrastructure
- SSD, up to 10s of GB/s, millions of IOPS, 100s PB of data
FSx for Lustre
- "Linux" && "Cluster"
- High Performanmce Computing (HPC)
- Can read S3 as file system through FSx, can write to S3
- Can be used from on-premise server
- One for hig burst, but temporary storage, no backup
- One fro long term storage, replicated in same AZ
- FTP related
- From/into S3 and EFS
Pre-processing data is Snowball Edge
HPC
A. No different as on premise solution, also bring extra management overhead for EC2
D.EFS, no directly way to create snapshot
E. File gateway doesn't support iSCSI
B. Never achieve data, so Glacier is not good
Choose C