FSx：Windows 文件服务器与 Lustre

什么是 Amazon FSx？

FSx for Windows File Server：概述

# Create an FSx for Windows file system joined to an AD domain
aws fsx create-file-system \
  --file-system-type WINDOWS \
  --storage-capacity 2000 \
  --storage-type SSD \
  --subnet-ids subnet-aaa111 subnet-bbb222 \
  --windows-configuration '{
    "ActiveDirectoryId": "d-1234567890",
    "ThroughputCapacity": 256,
    "DeploymentType": "MULTI_AZ_1",
    "AutomaticBackupRetentionDays": 7
  }'

Active Directory 集成

# Join FSx to a self-managed (on-premises) AD via Direct Connect
# aws fsx create-file-system with windows configuration:
# "SelfManagedActiveDirectoryConfiguration": {
#   "DomainName": "corp.example.com",
#   "OrganizationalUnitDistinguishedName": "OU=FSx,DC=corp,DC=example,DC=com",
#   "FileSystemAdministratorsGroup": "FSxAdmins",
#   "UserName": "svc-fsx@corp.example.com",
#   "Password": "...",
#   "DnsIps": ["10.0.0.10", "10.0.0.11"]
# }

FSx for Windows：部署类型

# Compare availability SLAs:
# Single-AZ 1 or 2: 99.9% SLA
# Multi-AZ 1: 99.99% SLA (synchronous standby, auto-failover)

# Access the FSx share from a Windows instance using UNC path:
# \\fs-0abc1234def567890.corp.example.com\share

# Or map a drive using PowerShell:
# New-SmbMapping -LocalPath 'Z:' -RemotePath '\\amznfsxabcd1234.corp.example.com\share'

FSx for Windows：备份和卷影副本

# Create an on-demand backup of an FSx Windows file system
aws fsx create-backup \
  --file-system-id fs-0abc1234def567890 \
  --tags Key=Purpose,Value=PreUpgradeBackup

# Configure shadow copy schedule via FSx console or PowerShell:
# $schedule = '{ "CopyFrequency": "DAILY_07_00" }'
# Set-FsxWindowsConfiguration -ShadowCopyConfiguration $schedule

# Shadow copies give end users self-service file recovery

FSx for Lustre：概述

# Create an FSx for Lustre file system linked to S3
aws fsx create-file-system \
  --file-system-type LUSTRE \
  --storage-capacity 1200 \
  --subnet-ids subnet-aaa111 \
  --lustre-configuration '{
    "ImportPath": "s3://my-hpc-data-bucket",
    "ExportPath": "s3://my-hpc-data-bucket/exports",
    "DeploymentType": "PERSISTENT_2",
    "PerUnitStorageThroughput": 250
  }'

FSx for Lustre：S3 集成

# Import data from S3 to FSx Lustre on demand
aws fsx create-data-repository-task \
  --type IMPORT_METADATA_AND_DATA \
  --paths 's3://my-hpc-data-bucket/dataset/' \
  --file-system-id fs-0abc1234def567890 \
  --report '{ "Enabled": true, "Path": "s3://my-hpc-data-bucket/task-reports/", "Scope": "FAILED_FILES_ONLY"}'

# Export results back to S3 after computation
aws fsx create-data-repository-task \
  --type EXPORT_TO_REPOSITORY \
  --paths '/results/' \
  --file-system-id fs-0abc1234def567890

FSx for Lustre：部署类型

# Scratch file system (no replication, cheaper, for temporary jobs)
aws fsx create-file-system \
  --file-system-type LUSTRE \
  --storage-capacity 1200 \
  --subnet-ids subnet-aaa111 \
  --lustre-configuration '{
    "DeploymentType": "SCRATCH_2",
    "ImportPath": "s3://my-data-bucket"
  }'

# Scratch 2 specs:
# 200 MB/s per TiB baseline
# 6x burst (data not replicated within AZ)

在 Linux 上挂载 FSx for Lustre

# Install Lustre client on Amazon Linux 2
sudo amazon-linux-extras install -y lustre

# Mount the FSx for Lustre file system
sudo mkdir /mnt/fsx
sudo mount -t lustre \
  -o relatime,flock \
  fs-0abc1234def567890.fsx.us-east-1.amazonaws.com@tcp:/fsx \
  /mnt/fsx

# Verify mount and check throughput
df -h /mnt/fsx
lsattr /mnt/fsx

FSx for NetApp ONTAP 和 OpenZFS

在 EFS、FSx for Windows 和 FSx for Lustre 之间进行选择

快速检查

课程回顾