vSAN Cache Hit Ratio: Diagnose and Fix Low Read Cache Hit Rate

Quick Answer

Adding a second NVMe cache device or upgrading to higher-endurance NVMe drives is correct because the metrics point directly at the cache tier running out of capacity during peak load, not at a networking or policy problem. In an all-flash vSAN disk group, the NVMe device serves purely as a write buffer and read cache accelerator sitting in front of the slower SATA SSD capacity tier; when the active working set during peak hours exceeds what that cache can hold, reads increasingly have to fall back to the capacity tier, which is exactly what a dropping cache hit ratio and rising read latency during a specific time window indicates. Since VDI logins and application launches during peak hours generate a burst of similar, repeated read activity, that is precisely the kind of workload that benefits from a larger or faster cache, and expanding the cache tier addresses the bottleneck without touching the storage policy, migrating VMs, or reconfiguring the cluster in any disruptive way. This differs from a scenario where latency is high evenly across all hours, which would point more toward an undersized capacity tier or network issues rather than a peak-hours cache problem. When performance degrades specifically during a predictable high-demand window and vSAN's own cache hit ratio metric drops alongside it, look for an answer that grows the cache tier rather than one that touches networking, policy, or compute.