Signal 1 — your SAN is aging, not your servers
When the storage layer is the refresh driver, HCI folds the SAN, its switches, and its support contracts into the node bill. One dual-controller mid-range SAN replacement typically funds a 3-node HCI cluster with comparable effective capacity — and removes a whole failure domain from the runbook.
Signal 2 — VM count is growing, change requests are not
If your team provisions VMs weekly but each request takes days of LUN carving and zone changes, the bottleneck is architecture, not effort. HCI provisioning is minutes and self-service friendly, which is the difference between a 2-person and a 5-person team at the same workload.
When to stay three-tier
Very large transactional databases, latency-critical cores, and workloads with predictable multi-petabyte growth still belong on dedicated storage. If your growth is steady rather than spiky and the SAN is under warranty, defer the switch — HCI economics thrive on refresh moments, not on mid-life migrations.
