VMware by Broadcom · Infrastruttura · YoctoIT tech page

vSAN

Hyperconverged storage: the data distributed across the nodes with per-VM policies — the SAN that disappears into the servers.

FOCUS · STORAGE IN THE NODESPolicies per VM, not per LUN: the storage administered from vCenter
Request the Free Assessment Learn more
YoctoIT material for clients and partners · VMware, vSphere, vSAN, NSX and the other products mentioned are trademarks of Broadcom Inc.
01 · What it is

vSAN, made clear.

vSAN aggregates the ESXi nodes' disks into a distributed datastore: no external SAN, local NVMe performance and — the real cultural novelty — storage policies per single VM: replicas, stripes and encryption declared as policy, enforced by the cluster. With the Express Storage Architecture, the efficiency is enterprise-array grade.

ESA
Express Storage Architecture: NVMe, compression and erasure coding without compromises
Per-VM
the policy on the workload, not on the LUN: each VM with its own protection level
1 console
storage administered from vCenter: the SAN trade simplified
vSAN
OFFICIAL VMWARE BRANDING · VSAN
CONSOLE REALE · VSAN DATA PROTECTION · FONTE: VMWARE BLOG
REAL CONSOLE · VSAN DATA PROTECTION · SOURCE: VMWARE BLOG
02 · How to use it well

The things that make the difference.

The cluster

The VMs and their policiesFTT, stripes, encryption declared
Node 1
Node 2
Node N (2-64)
local disks, distributed data
vSAN ESAthe software that makes the SAN
NVMe & 25/100GbEthe iron that's needed, right
The SAN, dissolved into the cluster

Failure domain design

FTT, host groups and network sized: resilience is designed before the first disk.

A network up to the task

vSAN lives off the network: 25GbE+ and a clean configuration or the performance lies.

Stretched cluster

Two sites in sync with a witness: geographic HA inside the platform.

Capacity and growth

You grow by nodes: capacity planning that accompanies, doesn't chase.

03 · In depth

ESA: the new storage architecture

vSAN Express Storage Architecture (ESA) eliminates the disk groups: all NVMe, log-structured filesystem, compression and erasure coding without penalties (RAID-5/6 at RAID-1's performance cost), fast native snapshots; the storage policies (SPBM) declare protection and performance per VM or per disk; the stretched cluster does RPO 0 between two sites with a witness; vSAN Max disaggregates the storage for the clusters that grow asymmetrically.

  • ESA — no disk groups, all NVMe: throughput and resilience doubled
  • SPBM — the per-VM policy: RAID, replicas and QoS declared, not hardwired
  • RAID-5/6 without the tax — erasure coding at mirroring's pace: capacity recovered
  • Stretched cluster — RPO 0 between two rooms with a witness: native metropolitan DR
  • Snapshot ESA — deep chains without degradation: frequent protection
  • vSAN Max — disaggregated storage: the compute clusters mounting an external pool
04 · Numbers and lifecycle

The numbers that matter.

~2x
the ESA vs OSA performance on the same hardware
RPO 0
with a stretched cluster
-40%
the typical TCO vs a traditional SAN
100%
policy-driven: the storage administered by intent
HCI takes the SAN out of the equation: sizing, policies and migration done by us — the storage that follows the VMs.
05 · Use cases

Where it really pays off.

New clusters

Hyperconvergence instead of the expiring SAN.

Sites and edge

Compact 2-3 node clusters, without external storage.

VDI and mixed workloads

Policies per profile: the desktop and the database, each its own.

Modern storage is software on the right nodes: vSAN sized and managed by us.