Broadcom repriced VMware while an 18 million dollar Oracle Database exposure hung over the same hosts. This case study shows how a UK media group contained both with a single cluster redesign, and why neither vendor set the scope for the other.
Oracle Database on VMware and a Broadcom subscription restructure hit the same estate at once, and handling them together avoided 18 million dollars of exposure.
This is the defining dual vendor problem of the current cycle. Broadcom repriced the platform, and the repricing forced infrastructure decisions that silently moved Oracle licensing boundaries. This case study shows how one UK media group refused to let either vendor set the scope for the other.
A UK media group discovered that its two largest infrastructure risks were the same risk. Oracle Database ran on VMware clusters, exposing the estate to a full cluster licensing position of about 18 million dollars, while Broadcom's move to subscription repriced the platform those same clusters ran on.
The group contained both in one coordinated move. A cluster redesign isolated Oracle onto a defined host group, the VCF renewal was right sized from the new layout, and no additional Oracle licenses were bought.
Each vendor priced the same physical hosts on a different logic. Broadcom billed the cores the clusters contained, while Oracle asserted every host its database could reach. Any change made to reduce one number moved the other, which is what makes the dual vendor squeeze different from two separate negotiations.
The complication was timing. The Broadcom renewal carried a date, the Oracle exposure did not, and a deadline on one side of a coupled problem pressures the whole estate into rushed decisions. Recognizing that asymmetry early shaped everything that followed.
Before the engagement, nobody did, which is typical. Infrastructure owned the VMware renewal, the database team owned Oracle deployment, and procurement owned two contracts it read separately. The first intervention was organizational: one owner, one estate model, and a rule that neither vendor received an answer derived from anything but that model.
Oracle treats VMware clusters as soft partitioning, so its audit position is that you license every host the database could ever run on. This stance flows from Oracle's partitioning policy, which is not contractual but is asserted in audits.
On a large cluster, that turns a handful of needed licenses into a full estate claim. For this media group, the gap was about 18 million dollars. The general mechanics are covered in our guide to Oracle Database licensing in virtualized environments.
Pinning Oracle workloads to defined hosts and isolating them in a dedicated cluster narrows the footprint. The wider Oracle Database licensing rules then apply only to that boundary, and the position is defensible when the isolation is documented and enforced.
A full cluster position is list arithmetic, not an invoice. Every host in asserted reach is counted, physical cores are converted through Oracle's core factor table, and the result is priced at Enterprise Edition and option list rates, with backdated support layered on top.
Two properties of that arithmetic matter to a buyer. Oracle's standard agreements contain no punitive multiplier, so the figure is a compliance gap priced at list, not a penalty. And every input, from host reach to option use, is contestable with evidence the customer controls.
Broadcom moved VMware from perpetual licenses to subscription bundles centered on VMware Cloud Foundation, priced per core. Standalone products were folded into the bundle, and renewals arrived carrying capacity many estates did not need.
For this media group the restructure raised cost sharply, and the renewal date fixed a deadline the Oracle question did not have. Public context on the acquisition is tracked on Broadcom news.
VMware perpetual versus Broadcom subscription
| Dimension | Perpetual era | Broadcom subscription |
|---|---|---|
| Licensing basis | Per socket perpetual | Per core subscription |
| Packaging | Standalone products | VCF bundle |
| Cost trajectory | Support renewals | Subscription uplift |
| Buyer lever | Maintain or drop support | Right size the core count |
The quoted VCF renewal is an opening position built from listed capacity, exactly as the Oracle figure is built from asserted reach. Three questions expose the gap: which cores actually run workloads, which bundle components have any deployment behind them, and what the same layout costs once the Oracle isolation is designed in.
Answering those questions before responding to the quote converts the renewal from a repricing event into a scoping negotiation. The group's answers came out of the same estate model the Oracle defense used, which is the entire point of building one.
Through infrastructure decisions. A VMware repricing forces choices about clusters, hosts, and hypervisors, and every one of those choices moves the boundary Oracle counts. The two bills are coupled through the architecture even though the vendors never appear in the same meeting.
Four coupling mechanisms did the damage across the estates we reviewed, and the first two were live in this engagement:
Four ways a Broadcom event moves the Oracle number
| Broadcom driven change | Oracle consequence |
|---|---|
| Consolidating clusters to cut subscription cores | Wider vMotion reach, more hosts inside Oracle's asserted scope |
| Refreshing to denser hosts with more cores | More licensable cores per host inside the Oracle boundary |
| Migrating workloads to another hypervisor or cloud | Every moved Oracle workload triggers a fresh licensing decision |
| Letting the VCF renewal deadline drive the timetable | Oracle decisions made under a clock Oracle did not even set |
The target architecture must be decided before either negotiation opens, because it is the input both vendors price. Decide the Oracle host group first, size the VCF core count from that layout, and only then talk numbers with anyone.
Run the sequence backward, by signing the VCF renewal first, and the consolidation baked into that renewal becomes an Oracle exposure you then discover. Our Oracle licensing under Broadcom VMware guide treats this interaction in full.
Broadcom renewals arrive with dates, and dates create pressure that spills onto decisions the renewal does not govern. This group deliberately decoupled the clocks: the VCF renewal got a negotiated timeline, and the Oracle containment work ran on its own schedule with no signature deadline at all.
The key decision was to treat Oracle and Broadcom as one problem on one estate. Isolating Oracle onto a dedicated cluster also shaped the VMware core count the group needed from VMware Cloud Foundation.
That single architectural change cut the Oracle footprint and reduced the Broadcom subscription scope at the same time. The work ran over a few months in three phases.
The starting artifact was a single inventory mapping every host, core, and Oracle workload across the clusters. Both exposures were priced from that one model: the full cluster Oracle position at about 18 million dollars, and the VCF renewal as quoted against cores the group did not fully use.
Putting both numbers on one page changed the internal conversation. Infrastructure, procurement, and licensing stopped optimizing separately, because the model showed every host decision moving both figures at once.
Oracle workloads moved onto a dedicated, license bounded cluster with documented affinity and isolation. The redesigned layout then defined the real VCF core requirement, which came in well below the quoted renewal capacity.
Only after the architecture was fixed did the negotiations open, each informed by the other. The Broadcom conversation priced the right sized core count, using the discipline in our Broadcom renewal response strategy. The Oracle conversation never became a purchase at all, because the contained footprint left nothing to sell against.
Handled separately, each vendor would have sized to a worst case. Handled together, the same hosts solved both problems, and neither vendor set the scope alone.
If the estate also runs Oracle middleware, the same coupling applies with a twist: WebLogic carries its own processor licensing and a Java SE entanglement, so a cluster redesign should check that boundary too. See the hidden Java SE coupling in WebLogic migrations before moving those workloads.
The common advice is to either buy enough Oracle licenses to cover the whole cluster or rip Oracle off VMware entirely. We disagree. In the Oracle on VMware cases we reviewed, both extremes cost far more than needed. Buying full cluster coverage pays for hosts the workload never touches, and a forced migration carries its own large bill. The buyer side move is architectural containment: isolate Oracle onto a defined, license bounded cluster so the footprint matches the workload. For this media group that approach contained roughly 18 million dollars of exposure without buying a single additional Oracle license.
Source: Redress Compliance advisory engagement file, 2024 to 2025.
We stopped treating Oracle and Broadcom as separate negotiations. One cluster redesign shaped both, and the eighteen million dollar Oracle exposure never turned into a purchase.
Roughly 18 million dollars of Oracle exposure contained without buying a single additional Oracle license, and a Broadcom subscription scoped to the cores the redesigned estate actually needed. The estate stayed on VMware Cloud Foundation, so the result cost no platform migration.
The containment rests on evidence, not assurance. The dedicated cluster, the affinity rules, and the change control records exist independently of any negotiation, so the position does not expire when an account team changes or a renewal comes back around.
The same records shaped the Broadcom side. A core count derived from a documented layout is difficult to argue upward, because the seller is no longer quoting against ambiguity.
An avoided license purchase also avoids the annual support stream that would have attached to it, and a right sized subscription resets the baseline every future renewal is quoted from. Both effects recur year after year, which is why the value of this close exceeds the headline exposure it contained.
Yes, when the containment is hard, documented, and current. Loose intentions do not hold; enforced boundaries with a paper trail do. The defensibility of the position rests on artifacts an auditor can test:
A formal audit would arrive from Oracle GLAS, formerly LMS, with 45 days written notice under the audit clause. The response would be the same evidence pack the containment produced, which is the point: the defense was built as a byproduct of the redesign, not assembled under audit pressure.
Containment does not depend on Oracle agreeing with it in advance. It depends on the customer being able to demonstrate, host by host, where the software can and cannot run, and on the contract saying nothing that surrenders the point.
If that letter ever arrives, the handling sequence, from the notice period through scoping and evidence exchange, is the standard one in our Oracle audit response playbook. The difference containment makes is that the playbook opens from a prepared position instead of a scramble.
Every VMware customer running Oracle will face some version of this squeeze before the current Broadcom cycle ends. Here is what to do next, in the order that preserves your leverage.
More buyer side outcomes, including the certification and Java matters this series covers, sit in our case study library.
Oracle treats VMware as soft partitioning, so its audit position is that you license every host the database could run on, including hosts vMotion could reach. The position rests on Oracle policy rather than contract, but it is asserted aggressively in audits.
Yes, with architecture. Pinning Oracle workloads to defined hosts and isolating them in a dedicated cluster narrows the licensable footprint to that boundary. Containment, documented and enforced, is what holds the position in an audit.
Roughly 18 million dollars. A full cluster licensing position would have required that spend. Architectural containment matched the Oracle footprint to the actual workload, so no additional Oracle licenses were purchased.
Broadcom moved VMware from perpetual licenses to per core subscription bundles built around VMware Cloud Foundation. Standalone products were folded into the bundle, which raised cost for many estates unless the core count was right sized.
Because the same hosts drove both problems. Isolating Oracle onto a dedicated cluster also shaped the VMware core count needed. Handled separately, each vendor would have sized to a worst case, so coordination avoided a double squeeze.
Neither, until the architecture is fixed. The target layout is the input both vendors price, so decide the Oracle host group and the resulting core count first. Then run the negotiations in parallel, letting each decision inform the other.
No. There is no obligation to notify Oracle of infrastructure changes, and volunteering them invites scrutiny on Oracle's timetable. Document the boundary internally with dated evidence, and present it if a formal audit, with its 45 days written notice, ever requires a response.
No. The estate stayed on VMware Cloud Foundation but with a redesigned layout and a right sized core count. The savings came from architecture and bundle discipline, not from a disruptive platform migration.
The architecture redesign and dual vendor negotiation ran over a few months. The sequencing mattered more than the speed, because each vendor decision informed the other and neither set the scope alone.
A buyer side framework for the VCF bundle, the per core conversion, and the Broadcom renewal cycle. Includes the core count worksheet, the bundle interpretation guide, and the renewal leverage map used across recent VMware engagements.
Independent. Buyer side. Built for CIOs and procurement leads carrying VMware estates moving onto Broadcom subscription terms. No vendor influence. No sales kickback.
Open the playbook in your browser. Corporate email only.
Open the Playbook →
We have served 500+ enterprise clients across 11 publishers. Every engagement starts with one conversation.
VCF bundle movement, per core subscription pricing, Oracle on VMware audit patterns, Broadcom renewal data, and the wider commercial leverage signals across every engagement.
