vSAN ESA in VCF 9.0: Planning and Deployment Guide

vSAN Express Storage Architecture (ESA) is the storage foundation for VCF 9.0 new deployments, and it is a significant architectural departure from the original vSAN OSA (Original Storage Architecture). If you are planning a VCF 9.0 deployment or evaluating a hardware refresh, understanding ESA's design principles, hardware requirements, and operational differences from OSA is essential before you commit to a bill of materials.

ESA vs. OSA: The Core Difference

The original vSAN architecture used a tiered model: fast cache devices (typically SSDs) accelerated reads and writes, while capacity devices (HDDs or SSDs) held the actual data. ESA eliminates this cache tier entirely. Instead, it requires all storage devices to be high-performance NVMe SSDs, and all of them serve as capacity devices simultaneously. This single-tier model reduces latency, simplifies the hardware design, and allows vSAN to apply compression and deduplication more efficiently without the overhead of managing cache flushing and destaging operations.

Hardware Requirements for ESA

ESA has stricter hardware requirements than OSA, which is the primary reason it is paired with hardware refresh projects. Each host requires a minimum of four NVMe SSDs with a capacity of at least 1.6 TB per drive. The drives must be on the VCF 9.0 Hardware Compatibility List (HCL) — not all NVMe SSDs qualify, and using non-certified drives is unsupported and will cause ESA to refuse to configure. Network requirements are also elevated: ESA performs best with 25 GbE or higher NICs, and 10 GbE deployments may see performance bottlenecks under heavy write workloads. Always validate your complete hardware bill of materials against the VCF 9.0 HCL before purchasing.

Storage Policies in ESA

vSAN uses VM Storage Policies to define how data is protected and how performance is allocated. ESA simplifies the policy model compared to OSA. The primary protection mechanism is RAID-5 or RAID-6 erasure coding for space efficiency, or RAID-1 mirroring for maximum performance. In ESA, RAID-5 erasure coding requires a minimum of four hosts and provides 1.33x space overhead compared to 3x for RAID-1 with three mirrors. RAID-6 requires six hosts and provides 1.5x overhead. For most production workloads, RAID-5 or RAID-6 is the right choice in ESA — the space savings over RAID-1 are substantial and the performance penalty is minimal on NVMe hardware.

Data Reduction: Compression and Deduplication

ESA delivers dramatically improved data reduction ratios compared to OSA. In OSA, compression and deduplication were applied after data was destaged from cache to capacity, which introduced latency and complexity. In ESA, compression is inline — data is compressed before it is written to the NVMe devices, which reduces write amplification and improves effective throughput. Deduplication in ESA operates at a block level across the entire cluster rather than per-disk-group, which yields significantly better dedup ratios for environments with substantial data similarity such as VDI, cloned VMs, or standardized application deployments.

Migrating from OSA to ESA

There is no in-place migration path from OSA to ESA. Moving from OSA to ESA requires standing up a new ESA cluster, migrating workloads via Storage vMotion or vSAN HCI Mesh, and decommissioning the OSA cluster. This is a planned migration project, not an upgrade operation, and it should be scoped accordingly. The good news is that VCF 9.0 fully supports running OSA clusters in existing workload domains during a phased transition, so you do not need to migrate everything at once. New workload domains in VCF 9.0 default to ESA when eligible hardware is detected.

Monitoring ESA Health

vSAN health checks in the vSphere Client and VCF Operations have been updated for ESA. Key health metrics to monitor include per-device wear indicator (NVMe SSDs have finite write endurance), resync operations (which occur after host failures or maintenance mode events), and capacity utilization including data reduction ratios. VCF Operations provides pre-built dashboards for vSAN ESA capacity planning and performance trending. Set alerting thresholds at 70% capacity utilization to give yourself adequate runway for expansion planning before hitting a critical threshold.

vSAN ESA represents a genuine architectural improvement that delivers real benefits for VCF 9.0 deployments. The hardware investment required is higher than OSA, but the performance, space efficiency, and operational simplicity gains justify the cost for most enterprise use cases. Plan your hardware refresh around ESA-certified NVMe devices, validate the HCL early, and right-size your storage policies for your workload mix.