Using the VMware Exit to Regain Control of Your Data Center
George Crump
George Crump is the founder of Storage Switzerland, an analyst firm covering storage, virtualization, and cloud infrastructure, and now works at VergeIO. He has advised enterprise IT leaders on data center architecture for more than 30 years, and he writes and speaks on the economics and control of modern infrastructure. His argument is direct. Infrastructure is too complex and brittle to support workloads of the future.
Today: the three clocks that took control of your data center, and what it takes to get all three back.
HCI or three tier, the same three clocks.
Renewal, refresh, and hardware, each set to someone else’s schedule. The VMware exit is your moment to take all three back.
- Every clock was set by a vendor, not by a workload.
- Control left one contract at a time, and each renewal hardens the terms.
- A buyer who cannot leave cannot negotiate.
The Renewal Clock
A vendor writes your renewal date, your bundle, and your terms. You just sign.
- Core-based licensing turns a processor choice into a math problem. The software bill routinely passes the hardware bill.
- Broadcom raised the minimum from 16 cores per CPU to 72, then customer pressure forced it back. The lever is still in the vendor’s hand.
Reported annual VMware cost for the same 10 servers, roughly 320 cores, before and after the 72-core minimum.
The Refresh Clock
A support calendar retires working servers, on the vendor’s schedule, not yours.
- Dell gives a PowerEdge generation about five years of standard support. HPE and Cisco run similar clocks.
- Enterprise servers run well for seven to ten years. The calendar forces the decision, not the equipment.
- Third-party maintenance survives as an entire industry because buyers want to keep good hardware running.
The Hardware Clock
A compatibility list decides which servers you are allowed to buy next.
- Step off the certified list and support goes with it. The vendor approves the hardware, not the buyer.
- vSAN licenses every server in the cluster by core, whether that node does storage work or not.
- The Nutanix controller VM reserves roughly 16 vCPUs and 64 GB per node before an application runs.
Four tiers become one OS.
Network, virtualization, storage, and data protection in a single codebase, on hardware you own.
- One install, one license, and one update cover the whole stack. No seams to certify.
- A running node evacuates and leaves the cluster with its workloads live, so the retirement date is yours.
- Generations of commodity x86 mix in one system, and storage scales without buying cores.
Take back all three clocks.
Renewal
One platform license. No core rounding, no repricing at the renewal date.
Refresh
Hardware runs until you retire it, not until a support calendar says so.
Hardware
Any commodity x86, generations mixed freely. No compatibility list.
The renewal date, the refresh timeline, and the hardware, back on your schedule.