06 — iSCSI Storage

Overview

Shared storage for the Failover Cluster is provided by hviscsi01 using the Windows Server iSCSI Target Server role — a free, built-in Windows Server feature that requires no additional licensing. The cluster nodes connect to the iSCSI target over two independent network paths (MPIO) on the dedicated Storage vSwitch.

Why iSCSI Over SMB for This Demo

Failover Clustering supports multiple shared storage types, but iSCSI is chosen here for the following reasons:

Consideration	iSCSI	SMB 3 (Scale-Out File Server)
Requires additional cluster	No — single server	Yes — requires a SOFS cluster
Windows built-in	Yes — no extra license	Yes — but requires extra VMs
MPIO support	Yes — dual paths	N/A (SMB multichannel)
Simulates real-world SAN	Yes	No
Demo complexity	Lower	Higher

The primary requirement is shared storage that Failover Clustering treats as a shared SCSI bus. iSCSI satisfies this and allows demonstrating MPIO path failover — a realistic enterprise scenario.

hviscsi01 VM Configuration

Property	Value
vCPU	4
RAM	16 GB
OS	Windows Server 2022
Mgmt IP	`172.16.10.11` (vSwitch-Mgmt)
Storage IP 1	`172.16.30.10` (vSwitch-Storage, path A)
Storage IP 2	`172.16.30.11` (vSwitch-Storage, path B)
VHDX (OS)	80 GB
VHDX (iSCSI data disk)	2 TB (thin provisioned)

The iSCSI data VHDX is kept separate from the OS VHDX. On demo day, move only the data VHDX to NVMe for performance.

iSCSI Target Server Role Installation

powershell

# On hviscsi01
Install-WindowsFeature -Name FS-iSCSITarget-Server -IncludeManagementTools

# Verify installation
Get-WindowsFeature -Name FS-iSCSITarget-Server

LUN Layout

LUN	Size	VHDX Name	Purpose
0	10 GB	`quorum.vhdx`	Cluster quorum disk (deprecated in favor of Cloud Witness, but useful for testing)
1	500 GB	`csv-vol1.vhdx`	Cluster Shared Volume 1 — workload VMs
2	500 GB	`csv-vol2.vhdx`	Cluster Shared Volume 2 — workload VMs
3	500 GB	`csv-vol3.vhdx`	Cluster Shared Volume 3 — workload VMs

The quorum disk is created for completeness but Cloud Witness is used as the actual quorum resource. See docs/08-cloud-witness.md.

Create the iSCSI Virtual Disks and Target

powershell

# On hviscsi01 — create the storage directory
New-Item -ItemType Directory -Path "E:\iSCSIVirtualDisks" -Force

# Create virtual disks (thin provisioned)
New-IscsiVirtualDisk `
  -Path "E:\iSCSIVirtualDisks\quorum.vhdx" `
  -Size 10GB

New-IscsiVirtualDisk `
  -Path "E:\iSCSIVirtualDisks\csv-vol1.vhdx" `
  -Size 500GB

New-IscsiVirtualDisk `
  -Path "E:\iSCSIVirtualDisks\csv-vol2.vhdx" `
  -Size 500GB

New-IscsiVirtualDisk `
  -Path "E:\iSCSIVirtualDisks\csv-vol3.vhdx" `
  -Size 500GB

# Create the iSCSI target
New-IscsiServerTarget `
  -TargetName "hvlab-clus01-target" `
  -InitiatorIds @(
      "IQN:iqn.1991-05.com.microsoft:hvnode01.azrl.mgmt",
      "IQN:iqn.1991-05.com.microsoft:hvnode02.azrl.mgmt",
      "IQN:iqn.1991-05.com.microsoft:hvnode03.azrl.mgmt",
      "IQN:iqn.1991-05.com.microsoft:hvnode04.azrl.mgmt"
  )

# Map virtual disks to the target
Add-IscsiVirtualDiskTargetMapping -TargetName "hvlab-clus01-target" `
  -Path "E:\iSCSIVirtualDisks\quorum.vhdx"

Add-IscsiVirtualDiskTargetMapping -TargetName "hvlab-clus01-target" `
  -Path "E:\iSCSIVirtualDisks\csv-vol1.vhdx"

Add-IscsiVirtualDiskTargetMapping -TargetName "hvlab-clus01-target" `
  -Path "E:\iSCSIVirtualDisks\csv-vol2.vhdx"

Add-IscsiVirtualDiskTargetMapping -TargetName "hvlab-clus01-target" `
  -Path "E:\iSCSIVirtualDisks\csv-vol3.vhdx"

# Verify
Get-IscsiServerTarget | Select-Object TargetName, IsConnected, InitiatorIds
Get-IscsiVirtualDisk | Select-Object Path, SizeBytes, IsAssigned

iSCSI Initiator Configuration on Cluster Nodes

Run the following on each of hvnode01 through hvnode04:

Enable and Start iSCSI Initiator Service

powershell

# On each cluster node
Start-Service MSiSCSI
Set-Service MSiSCSI -StartupType Automatic

Set the Initiator IQN

The IQN must match exactly what was specified in the target's InitiatorIds above:

powershell

# Check and set IQN (the default IQN is typically already correct)
$initiator = Get-InitiatorPort
Write-Host "Current IQN: $($initiator.NodeAddress)"

# If needed, set the IQN explicitly (rare — Windows sets this from hostname)
# The IQN should be: iqn.1991-05.com.microsoft:<hostname>.azrl.mgmt

Connect to iSCSI Target (Both Paths)

powershell

# Connect via path A (172.16.30.10)
New-IscsiTargetPortal `
  -TargetPortalAddress "172.16.30.10" `
  -InitiatorPortalAddress "172.16.30.21"  # Adjust last octet per node: .21, .22, .23, .24

Connect-IscsiTarget `
  -NodeAddress "iqn.1991-05.com.microsoft:hviscsi01-hvlab-clus01-target" `
  -TargetPortalAddress "172.16.30.10" `
  -InitiatorPortalAddress "172.16.30.21" `
  -IsPersistent $true

# Connect via path B (172.16.30.11) — second MPIO path
New-IscsiTargetPortal `
  -TargetPortalAddress "172.16.30.11" `
  -InitiatorPortalAddress "172.16.30.25"  # Adjust last octet per node: .25, .26, .27, .28

Connect-IscsiTarget `
  -NodeAddress "iqn.1991-05.com.microsoft:hviscsi01-hvlab-clus01-target" `
  -TargetPortalAddress "172.16.30.11" `
  -InitiatorPortalAddress "172.16.30.25" `
  -IsPersistent $true

Node Storage IP Assignments

Node	Path A IP	Path B IP
`hvnode01`	`172.16.30.21`	`172.16.30.25`
`hvnode02`	`172.16.30.22`	`172.16.30.26`
`hvnode03`	`172.16.30.23`	`172.16.30.27`
`hvnode04`	`172.16.30.24`	`172.16.30.28`

MPIO Configuration on Cluster Nodes

MPIO (Multipath I/O) must be installed and configured on every cluster node before running cluster validation:

powershell

# On each cluster node — install MPIO feature
Install-WindowsFeature -Name Multipath-IO -IncludeManagementTools
Restart-Computer -Force  # MPIO requires reboot

# After reboot — enable MPIO for iSCSI devices
Enable-MSDSMAutomaticClaim -BusType iSCSI

# Set the load balancing policy to Round Robin
Set-MSDSMGlobalDefaultLoadBalancePolicy -Policy RR

# Verify MPIO is claimed
Get-MSDSMAutomaticClaimSettings
Get-PhysicalDisk | Where-Object BusType -eq iSCSI | Select-Object FriendlyName, BusType

Verify MPIO Paths

After connecting both iSCSI paths, verify that MPIO sees two paths per disk:

powershell

# On each cluster node
Get-MSDSMSupportedHW
Get-Disk | Where-Object {$_.BusType -eq "iSCSI"} | ForEach-Object {
    $disk = $_
    Write-Host "Disk $($disk.Number): $($disk.FriendlyName)" -ForegroundColor Cyan
    Get-StorageReliabilityCounter -Disk $disk | Select-Object DeviceId
}

# Check MPIO path count (expect 2 per disk)
mpclaim -s -d

Disk Preparation on First Cluster Node Only

Initialize and format the disks on hvnode01 only. The other nodes will see the disks as cluster resources after cluster formation.

powershell

# On hvnode01 ONLY
# Get the iSCSI disks
$iscsiDisks = Get-Disk | Where-Object {
    $_.BusType -eq "iSCSI" -and $_.PartitionStyle -eq "RAW"
} | Sort-Object Size

# Initialize each disk (GPT for disks > 2 TB; MBR is fine for these)
foreach ($disk in $iscsiDisks) {
    Initialize-Disk -Number $disk.Number -PartitionStyle GPT -PassThru |
    New-Partition -AssignDriveLetter -UseMaximumSize |
    Format-Volume -FileSystem NTFS -AllocationUnitSize 65536 `
        -NewFileSystemLabel "iSCSI-$($disk.Number)" -Confirm:$false
    Write-Host "Initialized disk $($disk.Number) ($([math]::Round($disk.Size/1GB)) GB)"
}

After the Failover Cluster is created, these disks will be converted to Cluster Shared Volumes. Do not format them with any cluster-aware filesystem at this stage.

Firewall Rules on hviscsi01

powershell

# On hviscsi01 — ensure iSCSI Target firewall rules are enabled
Enable-NetFirewallRule -DisplayGroup "iSCSI Target (TCP-In)"
Enable-NetFirewallRule -DisplayGroup "iSCSI Target"

# Confirm
Get-NetFirewallRule | Where-Object DisplayGroup -like "*iSCSI*" |
    Select-Object DisplayName, Enabled, Direction

Verification Checklist

After completing iSCSI setup, verify:

powershell

# On hviscsi01 — confirm all 4 nodes are connected
Get-IscsiServerSession | Select-Object InitiatorNodeName, TargetName, IsConnected

# On each cluster node — confirm all 4 disks visible
Get-Disk | Where-Object BusType -eq "iSCSI" |
    Select-Object Number, FriendlyName, Size, HealthStatus

# Confirm MPIO — should show 2 active paths per disk
Get-MSDSMLoadBalancePolicy -Disk (Get-Disk | Where-Object BusType -eq "iSCSI")

Expected output: 4 iSCSI disks visible on each node, each with 2 MPIO paths, all showing HealthStatus: Healthy.

06 — iSCSI Storage ​

Overview ​

Why iSCSI Over SMB for This Demo ​

hviscsi01 VM Configuration ​

iSCSI Target Server Role Installation ​

LUN Layout ​

Create the iSCSI Virtual Disks and Target ​

iSCSI Initiator Configuration on Cluster Nodes ​

Enable and Start iSCSI Initiator Service ​

Set the Initiator IQN ​

Connect to iSCSI Target (Both Paths) ​

Node Storage IP Assignments ​

MPIO Configuration on Cluster Nodes ​

Verify MPIO Paths ​

Disk Preparation on First Cluster Node Only ​

Firewall Rules on hviscsi01 ​

Verification Checklist ​

06 — iSCSI Storage

Overview

Why iSCSI Over SMB for This Demo

hviscsi01 VM Configuration

iSCSI Target Server Role Installation

LUN Layout

Create the iSCSI Virtual Disks and Target

iSCSI Initiator Configuration on Cluster Nodes

Enable and Start iSCSI Initiator Service

Set the Initiator IQN

Connect to iSCSI Target (Both Paths)

Node Storage IP Assignments

MPIO Configuration on Cluster Nodes

Verify MPIO Paths

Disk Preparation on First Cluster Node Only

Firewall Rules on hviscsi01

Verification Checklist