Sample: VMware Renewal and Licensing Exposure Assessment
Sample deliverable. The client, environment, and all figures are fictional and for illustration only. Prepared by McSherry Tech for a fictional regional insurer with about 900 employees and 580 VMs.
1. Executive summary
The client is licensed for 768 VMware cores across two VCF 9.0 instances but runs on 560 physical cores. A third of the licensed cores, 256, sit in the two management domains, which run only VCF management components. Moving every cluster, management domains included, onto single-socket hosts before the renewal in 5 months cuts the licensed footprint to 528 cores, about $72,000 a year at an illustrative price. We recommend Option A for this renewal. Option B could save about another $108,000 a year, but only if Broadcom will quote vSphere Foundation and the security lead accepts replacing NSX segmentation; that decision is due within 30 days. Option C deserves a full review 9 months before the following renewal.
- Biggest finding: 208 licensed cores (27%) pay for capacity that does not exist, because every CPU is billed at a 16-core minimum.
- Management overhead: 8 of the 24 hosts and 256 of the 768 licensed cores are management domains. Both must stay at four hosts, so the saving comes from smaller hosts, not fewer.
- Deadline: Host orders must be placed within 30 days to finish consolidation before the renewal date.
- Compliance: All three options can meet NAIC and NIST CSF requirements. Option B needs replacement segmentation controls and new evidence; Option C carries the most audit risk in the next 12 months.
2. Scope and method
Three-week remote engagement. Data reviewed: an RVTools export from both sites, 30 days of vCenter performance data, the VCF Operations inventory of management components, the current Broadcom renewal quote, the hardware support contracts, and interviews with the infrastructure manager, the security lead, and the IT director.
3. Current state
The client runs VCF 9.0 as two VCF instances in one fleet: the primary site and a recovery site in another region. Each instance has its own four-host management domain, as VMware's multi-region design calls for. The management domains hold vCenter, NSX Manager, SDDC Manager, and the other management components for their instance, including the workload domain's own vCenter and NSX Managers. VCF Operations runs in the primary site's management domain.
| Cluster | Hosts | CPUs per host | Cores per CPU | Memory per host | Physical cores | Licensed cores | Runs | Avg CPU use | Avg memory use |
|---|---|---|---|---|---|---|---|---|---|
| Management domain, primary site | 4 | 2 | 12 | 512 GB | 96 | 128 | VCF management components | 9% | 21% |
| Production (workload domain) | 8 | 2 | 12 | 512 GB | 192 | 256 | 420 VMs | 22% | 61% |
| Dev/test (second cluster in the production workload domain) | 4 | 2 | 10 | 384 GB | 80 | 128 | 160 VMs | 18% | 47% |
| Management domain, recovery site | 4 | 2 | 12 | 512 GB | 96 | 128 | VCF management components | 6% | 15% |
| Disaster recovery (workload domain) | 4 | 2 | 12 | 512 GB | 96 | 128 | Replicas only | 6% | 18% |
| Total | 24 | 560 | 768 | 580 VMs |
The management domains use vSAN OSA. The workload clusters use a Fibre Channel SAN, so most of the vSAN capacity included with VCF goes unused. The NSX distributed firewall segments about 40 claims and payment VMs in production, and that segmentation is cited in the last NAIC-based exam. Production hosts reach end of hardware support in 7 months; the other hosts in 19 months.
4. Licensing exposure
VMware licensing is a per-core subscription. Every CPU is billed at a minimum of 16 cores (Broadcom KB 313548), and the count covers the physical cores on all ESX hosts in the environment, management domain hosts included. Renewing after the subscription anniversary can add a late-renewal penalty, reported at about 20% (ITdaily).
Reported per-core prices vary widely and rarely match a real quote, so this sample uses an illustrative $300 per core per year for VCF and $150 for vSphere Foundation. A real assessment uses the client's own quotes.
- Licensed today: 768 cores, about $230,400 a year
- Cores the hardware actually has: 560
- Cores paid for but not present: 208 (27%), about $62,400 a year
- Licensed cores in the two management domains: 256 (33%), about $76,800 a year, for hosts that run no business workloads
5. Options compared
| Measure | A. Stay on VCF 9, consolidate | B. Consolidate and change edition to vSphere Foundation | C. Move everything to Nutanix AHV |
|---|---|---|---|
| Licensed VMware cores | 528, including 192 in the two management domains | 336 (vSphere Foundation); no management domains | 0 after one bridge renewal |
| Annual platform cost (illustrative) | $158,400 | $50,400 | $90,000 to $120,000 Nutanix |
| One-time cost (illustrative) | 20 hosts: 12 for workloads, 8 for the management domains. Production replacement is already due in 7 months. | 12 workload hosts, replacement segmentation controls, and rebuilding both sites as vSphere Foundation | $150,000 to $250,000 migration, training, and services |
| Time to complete | 3 to 4 months | 4 to 6 months | 9 to 12 months |
| Main risk | DR sized for critical systems only; management hosts swapped one at a time | Loses the NSX distributed firewall and VCF lifecycle tooling; segmentation evidence must be rebuilt | Audit evidence and staff skills during migration |
| Fits the 5-month renewal? | Yes, if hosts are ordered by day 30 | Only if quoted and approved by day 30, with a short bridge if the rebuild runs long | No |
Option A design, workload clusters. Production on 6 single-socket 32-core hosts with 768 GB of memory each (192 cores, 4,608 GB). Production memory use of about 2,500 GB averages 65% with one host held in reserve. Dev/test on 3 single-socket 24-core hosts with 512 GB each (72 cores), at about 71% memory with one host in reserve. DR on 3 single-socket 24-core hosts with 768 GB each (72 cores), sized to run the critical tier only.
Option A design, management domains. Both sites keep four-host management domains, moved to single-socket 24-core hosts with 512 GB each (96 cores per site instead of 128). Four hosts is the minimum for a high-availability management domain, so the hosts get smaller rather than fewer. The primary site runs Broadcom's High Availability, Small management profile (106 vCPU, 335 GB of memory, and 8,422 GB of disk) plus the production workload domain's three NSX Managers at 6 vCPU and 24 GB each, which comes to 124 vCPU and 407 GB before the workload domain's vCenter. With one host in reserve, that is under 2 vCPU per physical core and under 30% of memory, which leaves room for that vCenter. The recovery site runs a smaller set, with no VCF Operations or VCF Automation of its own. The new hosts must come from the same server vendor as today's management hosts, because VCF requires a single vendor in the management domain, and must be certified for vSAN OSA so they can join the existing management clusters and replace the old hosts one at a time.
Option B design. The same 12 workload hosts as Option A. vSphere Foundation does not use NSX, SDDC Manager, or VCF Automation, so in this design vCenter and VCF Operations run inside the production and DR clusters and both four-host management domains are retired. Production memory rises to about 75% at peak with one host in reserve.
6. Risk register
| # | Risk | Likelihood | Impact | Mitigation |
|---|---|---|---|---|
| 1 | Host lead times push consolidation past the renewal date | Medium | High | Order all 20 hosts within 30 days; ask Broadcom for a short co-termed bridge if needed |
| 2 | Smaller DR cluster cannot run all production systems | Medium | High | Document a tiered recovery plan; test critical systems failover quarterly |
| 3 | The management cluster image can't cover both the old and new server models during the swap | Medium | Medium | Validate the vSphere Lifecycle Manager image and vendor add-on for both models before the first host is replaced |
| 4 | New management services later outgrow the right-sized management domain | Low | Medium | Before adding VCF Automation or other management services, check capacity against Broadcom's High Availability, Medium profile (184 vCPU, 656 GB) |
| 5 | Core count reported at renewal does not match deployed hardware | Low | High | True-up the core inventory from RVTools at both sites, management domains included, before submitting the renewal |
| 6 | Broadcom does not quote vSphere Foundation, or quotes it after the host order deadline | Medium | Medium | Request VCF and vSphere Foundation quotes in week 1; decide on Option B by day 30 |
| 7 | Staff lack Nutanix skills for Option C | High | Medium | Fund Nutanix training for two engineers before the next renewal decision |
7. Recommendation and 90-day roadmap
Choose Option A for this renewal, unless both Option B conditions are met by day 30: a vSphere Foundation quote in hand and the security lead's written acceptance of replacement segmentation controls. Revisit Option C 9 months before the following renewal.
| Days | Actions | Owner |
|---|---|---|
| 1 to 30 | Confirm the renewal date. Request VCF and vSphere Foundation quotes. Approve and order 20 hosts: 6 production, 3 dev/test, 3 DR, and 4 management hosts per site. Validate the management cluster image for the new model. Decide on Option B with the security lead. Freeze new VM growth outside the plan. | IT director |
| 31 to 60 | Build the consolidated workload clusters. Migrate VMs with vMotion. Replace management domain hosts one at a time, recovery site first. Retire the old hosts. | Infrastructure manager |
| 61 to 90 | True-up the core inventory at both sites. Submit the renewal at 528 cores. Run the first tiered DR test. Schedule the Option C review. | Infrastructure manager and security lead |
Appendix: assumptions
- Illustrative prices of $300 per core per year for VCF and $150 for vSphere Foundation; replace with the client's quotes. Whether vSphere Foundation is offered at renewal is Broadcom's decision.
- Nutanix costs are fictional ranges for illustration; a real assessment models them at planning level from the actual workloads.
- Management domain sizing uses Broadcom's High Availability, Small deployment profile for the primary site. A real assessment sizes from the client's VCF Operations inventory of deployed components.
- Four hosts per management domain is the high-availability minimum. VCF also supports a minimal-footprint model that runs management components and workloads in one cluster; it trades away separation and availability, so it was not considered for a regulated insurer.
- Production peak CPU use of 45% fits on 192 cores with one host held in reserve.
- Production peak memory use of 68% (about 2,785 GB) runs at about 73% on the 5 hosts left when one is held in reserve.