LMS Scripts and Audit Data

How Scripts Overcount in Virtualized Estates

Oracle scripts overcount in virtualized estates because they read every host the software could reach, and with VMware, Hyper V or KVM not recognised as hard partitioning, the count spreads across the whole cluster. Reconciled against the real boundary, a line by line review of virtualization findings typically cuts the claim by 60 to 80 percent.

Why do Oracle scripts overcount in virtualized estates?

Oracle scripts overcount in virtualized estates because they read every host the software could theoretically reach rather than the hosts that actually run it. In a shared hypervisor cluster, the collection tools gather data across layers, and the resulting count assumes the workload could move to any host, so every core in the cluster becomes licensable in the raw output. The number is large by design, and it lands before any human has reconciled it against your real deployment. This is the most common reason a virtualization finding arrives looking far worse than the truth. The full treatment is in the Oracle Virtualization Licensing Guide.

Does Oracle recognise VMware as hard partitioning?

No, Oracle's partitioning policy does not recognise VMware, Hyper V or KVM as hard partitioning, and that single fact is what lets the scripts assume a cluster wide boundary. Because the policy treats these hypervisors as soft partitioning, Oracle's position is that the software could run anywhere in the cluster and so must be licensed everywhere in it. The crucial point for buyers is that this is a policy stance, not a contract term, and contract language beats policy. The script output reflects the policy view, which is the widest possible reading, not necessarily the one your signed agreement supports.

The buyer move

Map the real boundary. Identify the hosts that genuinely run Oracle, document the controls that keep the workload there, and reconcile that against the contract before the cluster wide number becomes the agreed number.

How does counting across layers inflate the number?

Counting across layers inflates the number because the same physical capacity can be read more than once when the tools traverse the virtualization stack. A host counted at the cluster level, then again through a management layer, produces a total that exceeds the real hardware. Add the cluster wide assumption on top, and the raw figure can be a multiple of what the deployment actually consumes. None of this is malicious, it is what the tools do when pointed at a complex estate, but it is why the output is a draft and never a verdict.

Where the overcount comes from and how to answer it.
Source of overcountWhat the script assumesBuyer move
Cluster wide reachSoftware could run on any hostProve the real boundary
Layered countingSame capacity seen twiceDeduplicate the hosts
Core factor ignoredRaw cores, not adjustedApply the core factor table
Policy as contractPartitioning policy governsRead the signed agreement

How do you reconcile the script output?

You reconcile the script output by rebuilding the count from the real deployment up, host by host, and applying the contract and the core factor table at each step. Start from the hosts that actually run Oracle, remove the double counting introduced by the layers, apply the correct core factor, and test the boundary against what your agreement says rather than what the policy asserts. The gap between the raw figure and the reconciled figure is the heart of the negotiation, and in a virtualized estate it is often very large. For the data discipline behind this, see feature usage data, the DBA views that matter and counting processors correctly with core factors.

What is the next step?

The next step is to make sure no virtualization script output reaches Oracle until it has been reconciled against your real boundary and your contract. A cluster wide number submitted raw is the single most expensive mistake in a virtualized audit, and it is entirely avoidable. The Oracle Audit Defense Handbook gives you the boundary mapping framework and the reconciliation method to bring the figure back to the truth.

Next step

Download the Oracle Audit Defense Handbook for the virtualization boundary checklist and the script reconciliation method. Get it from the white paper library.

FAQ

Questions buyers ask.

Because the scripts read every host the software could reach. With VMware, Hyper V or KVM not recognised as hard partitioning, the tools count cores across the whole cluster rather than the hosts that actually run Oracle.
No. Oracle's partitioning policy does not recognise VMware, Hyper V or KVM as hard partitioning, which is why script output assumes a cluster wide boundary. That assumption rests on policy, and contract language beats policy.
Yes. Reconcile the script output against the real deployment boundary and your contract before submission. A line by line review of virtualization findings typically cuts the claim by 60 to 80 percent.
Download guide

Get the Oracle Audit Defense Handbook.

The full buyer side framework for mapping the virtualization boundary, reconciling script output and running the line by line review that cuts findings 60 to 80 percent. Free with a work email.

The License Position, our weekly buyer side note, lands in your inbox when you download. New York and London. We never publish a public email address.