Symptoms
Virtual machine performance degradation, disk queuing, or increased I/O wait times.
vCenter performance charts showing spikes in "Device Latency" or "Physical Disk Latency."
Purpose
This article provides guidance on healthy latency ranges for vSAN devices (NVMe, SSD, and Magnetic) and explains how these correlate with industry-standard health thresholds.
VSAN 8.x, VSAN 9.x
If the vSAN drives exceed the recommended thresholds, investigate the following:
Controller Queue Depth: An overloaded storage controller cannot process I/O fast enough, causing "backlog" latency.
Firmware Mismatch: Outdated firmware on HBAs or SSDs can cause inconsistent performance.
vSAN Re-sync: If the cluster is rebuilding data after a failure, background traffic will naturally increase disk latency
Latency is the time it takes for a storage device to complete a single I/O request. In vSAN, the "healthy" range varies drastically depending on the physical hardware being used.
Vsan Expected Latency: < 500 microseconds (0.5 ms)
Characteristics: NVMe communicates directly with the CPU via the PCIe bus, bypassing the overhead of traditional SATA/SAS controllers.
Healthy Range: Consistent sub-millisecond response times are expected even under moderate load. Latency consistently above 1ms in an NVMe-backed cluster typically indicates bus congestion or driver/firmware issues.
vSAN Expected Latency: Up to 1 ms
Characteristics: Standard enterprise flash drives are significantly faster than spinning disks but are limited by the SATA or SAS protocol overhead.
Healthy Range: Generally stays between 0.5ms and 1.5ms. If latency consistently exceeds 2-3ms, it often indicates the drive is reaching its "write cliff" or that there is high contention at the controller level.
vSAN Expected Latency: 10 ms to 20 ms
Characteristics: Performance is physically limited by the seek time (the time for the head to move) and rotational latency (the time for the platter to spin).
Healthy Range: 15ms-20ms is considered a "normal" working state for a 7.2K or 10K RPM drive. Latency above 30ms is a critical warning sign that the disk is overwhelmed (I/O queuing).
The following table correlates vSAN monitoring values with industry-standard benchmarks for "healthy" storage performance.
| Media Type | vSAN Healthy Target | Typical Industry Standard Alert Threshold | Performance State |
| NVMe | < 0.5 ms | > 1 ms | High-performance / Ultra-low latency |
| Enterprise SSD | < 1 ms | > 3 ms | Standard All-Flash Performance |
| 10K/15K HDD | 5 ms – 15 ms | > 25 ms | Performance Hybrid Tier |