Oracle Exadata is the engineered system that combines Oracle database servers, storage servers, and a high speed internal network into a single platform optimised for the Oracle database. Exadata is available as an on premises appliance, as Exadata Cloud at Customer where Oracle owns and operates the hardware in the buyer data centre, and as Exadata Cloud Service in the Oracle public cloud. Each consumption model carries a different commercial structure and a different set of negotiation levers. The common feature across all three is the bundle that combines hardware, database licences, and options into a single quote that obscures the component pricing.
The bundle problem.
The Exadata quote is presented as a single platform price that combines the engineered hardware, the database Enterprise Edition licences for the cores in the system, and a set of mandatory and optional database options. The bundle is presented as a discounted package. The single price obscures the net price of each component and prevents the buyer from benchmarking the components against the market.
The negotiation depends on unbundling the quote into its components. The buyer should require a line item breakdown that shows the hardware price, the database licence price per processor, and the price of each option. The line item breakdown exposes the components that are negotiable and the components that are inflated. See the Enterprise Edition discount note for the licence component benchmarking.
The hardware component.
The hardware component of the Exadata quote is the engineered system itself. The hardware is priced by configuration size from the smallest eighth rack to the full rack and beyond. The hardware price is negotiable on the same timing drivers as the database licences and is most negotiable at the Oracle quarter end and fiscal year end.
The hardware component also carries a hardware support stream that is separate from the database software support. The hardware support covers the engineered system maintenance and is renewed annually. The buyer should benchmark the hardware support rate and should negotiate the hardware support alongside the hardware purchase rather than accepting the standard rate.
The licence component.
The licence component is the database Enterprise Edition licences required for the processor cores in the Exadata system. The Exadata core count is high and the licence component is frequently the largest part of the total quote. The licence is calculated on the core count multiplied by the Oracle core factor. The buyer should establish the exact core count being licensed and should confirm that the licence count matches the cores that will actually be enabled.
Exadata supports capacity on demand which allows the buyer to enable a subset of the cores and to license only the enabled cores. The capacity on demand feature reduces the licence component substantially where the workload does not require the full core count. The buyer should size the enabled cores to the actual workload and should license only the enabled cores. See the processor licence audit note for the core counting detail.
The Cloud at Customer model.
Exadata Cloud at Customer places the Oracle owned hardware in the buyer data centre and charges on a consumption basis rather than on a perpetual licence basis. The model converts the capital expense of the hardware and licences into an operating expense paid through Oracle cloud credits. The consumption model carries its own negotiation levers around the credit rate, the committed consumption, and the term.
The Cloud at Customer model is attractive where the buyer wants to remove the data centre hardware ownership while keeping the data on premises for sovereignty or latency reasons. The buyer should negotiate the committed consumption carefully because an over committed consumption obligation becomes a stranded cost if the workload does not grow as forecast. See the Autonomous Database pricing note for the consumption model mechanics.
The lock in consideration.
The Exadata platform creates a degree of technical lock in because the database features that exploit the engineered system are specific to Exadata. A buyer that adopts Exadata specific features such as smart scan and storage indexes will find that the workload does not migrate cleanly to a generic platform. The lock in increases the Oracle leverage at the next renewal.
The buyer should weigh the performance benefit of the Exadata specific features against the lock in they create. A buyer that wants to preserve the option to migrate should limit the dependence on Exadata specific features. See the AWS RDS migration note for the alternative platform path.
The negotiation sequence.
The Exadata negotiation should proceed by unbundling the quote, benchmarking each component, sizing the enabled cores to the workload, and assembling a counter offer on each component. The counter offer should address the hardware net price, the licence net price per processor, the option scope, and the support rate. The counter offer should be timed to the Oracle fiscal year end where the discount is deepest.
For the wider cluster see Database Negotiation. For the service see New License Procurement. For the deal structure see Database Licensing. For the Oracle product see Oracle Database. For the full research read the Oracle Negotiation Playbook.
Engaging an independent advisor.
The Exadata negotiation benefits from independent benchmarking of each component and from independent sizing of the enabled core count. An independent advisor holds component pricing across comparable Exadata deals and can establish whether the hardware, licence, and option components are competitive. The advisor can also model the capacity on demand sizing and the Cloud at Customer consumption commitment to avoid the stranded cost trap.
A North American bank engaged an advisor before a full rack Exadata purchase in 2024. The advisor unbundled the quote, established that the licence component was sized to the full core count rather than the workload, and resized the purchase to capacity on demand on approximately sixty percent of the cores. The resizing and the component benchmarking reduced the total deal value by approximately three point two million dollars.