Setting Up Turnkey Distributed Storage with vSAN for 3 Nodes

Setting Up Turnkey Distributed Storage with vSAN for 3 Nodes

Imagine three servers with local SSDs, each 10 TB. The task is to combine them into a single fault-tolerant datastore without buying an external SAN. vSAN solves this, but without proper configuration, you risk getting a slow virtual machine or a never-ending resync. We have configured vSAN for 40+ projects—from a 3-node cluster for small businesses to a Stretch Cluster with quorum across two sites. Our engineers are VMware Certified Professionals (VCP), guaranteeing smooth deployment and 30 days of post-go-live support.

vSAN turns local disks of ESXi hosts into a single cluster datastore. No external arrays, no SAN. Each host simultaneously consumes storage and provides its disks to the common pool. As stated in VMware vSAN Planning and Deployment Guide, 'A minimum of three hosts is required for vSAN fault tolerance.' This is convenient until you need to understand why erasure coding overhead impacts performance or why after replacing a disk, resync took two days.

Why vSAN Outperforms Traditional SAN?

Compare: SAN requires two switches, two controllers, a dedicated LUN, and admin time for zoning. vSAN is all-in-one: local disks + 10GbE network. In our tests, vSAN All-Flash delivers <1 ms latency for random 4K writes, which is 30% better than comparable entry-level SAN. Storage cost is 40% lower due to the absence of external arrays. For a typical 3-node all-flash cluster with 10 TB usable, total cost including licensing and hardware upgrades starts at approximately $12,000.

Prerequisites and Planning

vSAN requires a minimum of 3 hosts with vSphere Enterprise Plus or vSAN license. Fewer than three nodes—no quorum, no fault tolerance. Each host needs at least two types of disks.

All-Flash vs Hybrid: What to Choose?

All-Flash is the standard for new deployments. Hybrid remains only where HDDs already exist and replacing them is economically unfeasible.

Network requirements: dedicated VMkernel network for vSAN traffic, minimum 10 Gbps. On gigabit networks, vSAN works, but resync after disk replacement will take hours instead of minutes. For advanced configurations, consider vSAN iSCSI service or witness node for Stretch Cluster.

vSAN Configuration: What's Included

We perform a turnkey project:

• Audit of current infrastructure (compatibility, network, disks)
• Design: cluster, disk groups, Storage Policies (including RAID-5/6 erasure coding and FTT)
• Deployment: enable vSAN, configure deduplication, optimization of I/O path
• Documentation: cluster diagram, Storage Policies, maintenance runbook
• Training: how to monitor, change disks, respond to alerts
• 30 days of guaranteed support after go-live

We will assess your project in 1 day, implement in 3–10 days. Contact us for a consultation—we will select the optimal configuration for your tasks.

Enabling vSAN via vSphere Client

The process is GUI-based, but key decisions must be made in advance:

1. Disk group configuration—no more than 7 capacity disks per one cache disk.
2. Fault tolerance mode via Storage Policy: FTT=1 RAID-1 (2 copies), FTT=1 RAID-5 (4 hosts), FTT=2 RAID-6 (6 hosts).
3. Failures To Tolerate—for example, for three nodes only FTT=1 is possible.
4. Deduplication and compression—only on All-Flash, enable during a maintenance window.

Enabling via PowerCLI for automation:

# Enable vSAN on the cluster
$cluster = Get-Cluster -Name "Production-Cluster"
Set-VsanClusterConfiguration -Configuration $cluster -SdrsAutomationLevel FullyAutomated

# Create Storage Policy
New-SpbmStoragePolicy -Name "vSAN-FTT1-RAID5" `
    -Description "RAID-5, FTT=1, 4 hosts required" `
    -RuleSets @(
        New-SpbmRuleSet -AllOfRules @(
            New-SpbmRule -Capability (Get-SpbmCapability -Name "VSAN.hostFailuresToTolerate") -Value 1,
            New-SpbmRule -Capability (Get-SpbmCapability -Name "VSAN.replicaPreference") -Value "RAID-5/6 (Erasure Coding) - Capacity"
        )
    )

Storage Policy—the Heart of vSAN

The key difference from traditional SAN: you do not format LUNs and assign VMs to a datastore with the desired protection level. In vSAN, you apply a Storage Policy to the virtual machine, and vSAN automatically places data according to the policy.

Critical detail: a mismatch between the policy and the current cluster state is displayed as Non-Compliant. For example, a policy requires FTT=2, but the cluster has only 3 hosts—vSAN cannot meet the requirement and will show non-compliance. Data remains accessible but without the required protection level.

Check compliance via CLI:

Get-VM | Get-HardDisk | Get-SpbmEntityConfiguration | Where-Object {$_.ComplianceStatus -ne "compliant"} 

How to Improve vSAN Performance?

Key levers:

• Switching to All-Flash reduces latency by 5x compared to Hybrid.
• Enabling deduplication saves 30-50% on typical VDI workloads.
• Use a dedicated cache disk: no more than 7 capacity disks per one NVMe.
• If cache congestion occurs—scale the number of cache disks or switch to NVMe.
• Consider vSAN object recalculation for better load balancing.

Monitoring and Diagnostics

Proactive monitoring via vsan.disks_stats in esxcli:

# Via SSH on the ESXi host
esxcli vsan storage list
esxcli vsan health cluster list
# Object component status
esxcli vsan debug object list | grep -E 'State|Policy'

Key metrics in vSphere Client → vSAN → Monitoring:

• Resync Throughput and Resync ETL—when replacing a disk, we monitor that resync completes before the next possible failure.
• Congestion—if above zero, the cache disk is overloaded. Time to add a cache disk or switch to NVMe.
• Disk latency—>5ms for capacity tier on All-Flash indicates a disk problem.

Alert on degraded component via vRealize Operations or SNMP trap—mandatory. A silent Absent component status is worse than Degraded: data exists, but protection is already gone.

Common Issues

• VM loads slowly after migration—check cache congestion on the destination host.
• Resync does not complete—limit the speed via esxcli vsan resync schedule --rate-limit (100% at night, 50% during the day).
• Free space runs out unexpectedly—overhead from RAID and metadata can be up to 30%; plan usage no higher than 70%.
• After disk replacement, resync takes a day—ensure the network is 10GbE and there are no other I/O loads.

How Much Does vSAN Configuration Cost?

Cost is calculated individually after an audit. For a typical 3-node all-flash cluster with 10 TB usable, total cost including licensing and hardware upgrades starts at approximately $12,000. We have been doing VMware infrastructure for 7 years and have implemented 40+ vSAN projects. Our VCP-certified engineers guarantee smooth migration and 30 days of support. Contact us—we will assess your task in 1 day.

Timelines

VMware vSAN license required. We use official VMware documentation vSAN Planning and Deployment Guide and years of experience. Get a consultation—just write to us.