Cloud at Customer vs Dedicated Region. The licensing read.

Oracle ships two on premise OCI paths. Cloud at Customer is a single rack appliance. Dedicated Region is a full OCI region installed inside the customer four walls. The two carry different price floors, different licensing rules, and different audit postures.

Cloud at Customer starts around $500K per year on the smallest configuration. Dedicated Region starts around $6M per year on the minimum twelve rack footprint. The buyer side question is which footprint the running workload actually needs.

Read this alongside the Cloud at Customer guide, the Dedicated Region guide, the Oracle knowledge hub, the Oracle advisory practice, and the Vendor Shield subscription.

Cloud at Customer scope

Cloud at Customer is the smaller of the two paths. The footprint is a single Oracle Cloud appliance rack installed inside the customer data center. The unit is metered per OCPU hour at the same rate as public OCI compute.

Shape and capacity

• Single rack form factor. Around 38 to 50 OCPUs per rack on the standard configuration.
• Curated OCI catalog. Compute, block storage, Autonomous Database, and Exadata services are available. Many higher tier services are not.
• Oracle managed. Oracle staff hold root access. The customer holds workload control.
• Four year minimum term. The contract floor commit carries a four year horizon.

Cloud at Customer price bands

Configuration	OCPUs included	Annual floor commit	Best fit
Quarter rack	32	$500K to $700K	Single workload pilot
Half rack	64	$900K to $1.3M	Mid sized data residency case
Full rack	128	$1.6M to $2.4M	Production on premise OCI
Two rack cluster	256	$2.8M to $4.2M	Multi workload production

Dedicated Region scope

Dedicated Region is the larger path. The footprint is a full OCI region installed inside the customer data center. The minimum order is twelve racks. The price floor sits around six million dollars per year.

Shape and capacity

• Twelve rack minimum. The smallest Dedicated Region carries roughly 1,500 OCPUs on the entry shape.
• Full OCI catalog. Every public OCI service is available. The catalog mirrors a public region.
• Oracle managed control plane. Oracle handles patching, hardware, and operating system.
• Four year minimum term. Same floor commit horizon as Cloud at Customer.

Dedicated Region price bands

Footprint	OCPUs included	Annual floor commit	Best fit
Entry twelve racks	1,500	$6M to $8M	Single regulated estate
Mid sized twenty four racks	3,000	$10M to $14M	Banking primary region
Large forty rack	5,000	$16M to $22M	National data sovereignty
Multi region	10,000 plus	$30M plus	Sovereign cloud anchor

Cost and floor commit

The two paths sit at very different price floors. The buyer side response is to model the workload demand against the floor before any internal champion sign off.

Three cost variables to model

1. Workload OCPU demand. Sum the peak demand across every workload that the on premise OCI is expected to host.
2. Service catalog need. List every higher tier service that the workload calls. Cloud at Customer may not cover it.
3. Floor commit utilization. The contract burns the floor in any case. Model the utilization curve across the four year term.

Floor commit burn comparison

Year	CaC quarter rack	DDR entry footprint	Utilization note
Year one	$500K	$6M	Typical sixty percent utilization
Year two	$525K	$6.3M	Five percent escalator on each
Year three	$551K	$6.6M	Eighty percent utilization if on track
Year four	$579K	$6.95M	Renewal optionality before year five

Licensing posture

Both paths accept Oracle Database, Middleware, and Java BYOL. The buyer side response is to apply existing entitlements before paying for license included rates.

BYOL math on both paths

• Two OCPUs equals one processor on x86. The core factor table applies on both Cloud at Customer and Dedicated Region.
• One OCPU equals one processor on Arm Ampere. Apply the Arm core factor where the workload supports it.
• Java SE bundled inside WebLogic. The bundled Java SE entitlement covers WebLogic JVMs on either path.
• Enterprise Edition mandatory on Exadata. Cloud at Customer Exadata service requires Database Enterprise Edition entitlements.

BYOL savings comparison

Path	License included rate	BYOL rate	Annual saving on 100 OCPU base
Cloud at Customer x86	$3.06 per OCPU hour	$1.27 per OCPU hour	Around $1.57M
Cloud at Customer Arm	$1.27 per OCPU hour	$0.40 per OCPU hour	Around $760K
Dedicated Region x86	$3.06 per OCPU hour	$1.27 per OCPU hour	Around $1.57M
Dedicated Region Arm	$1.27 per OCPU hour	$0.40 per OCPU hour	Around $760K

Partition policy treatment

Oracle still applies the partition policy on both on premise OCI paths. The buyer side response is to architect the cluster around the approved technologies and document the BYOL count.

Three partition rules to know

1. OCPU partitioning is approved. Oracle accepts OCPU level partitioning on both Cloud at Customer and Dedicated Region.
2. VMware on the racks is not approved. The customer cannot drop VMware vSphere on either path.
3. Dedicated Region accepts third party hypervisors only on specific shapes. Approval requires an explicit Oracle written confirmation.

Audit posture by path

Path	BYOL count owner	Audit frequency	Friendly or strict
Cloud at Customer	Customer	Annual review	Friendly when documented
Dedicated Region	Customer	Annual review	Friendly when documented
OCI public region	Oracle	Continuous	Friendly
AWS or Azure BYOL	Customer	Triennial	Strict

Seven negotiation levers

The buyer side has seven specific levers on either path. Each maps to one cost line or one risk line.

Seven levers worth pursuing

1. Negotiate the floor commit ramp. Insert a year one and year two reduced floor.
2. Shorten the term. Push from four years to three where Oracle will accept it.
3. Expand the service inclusion. On Cloud at Customer, push to add Functions and OKE without the Dedicated Region price step.
4. Cap the annual escalator. Hold the year over year increase at three percent or below.
5. Insert a partial exit clause. Allow scope reduction at the half term point.
6. Apply BYOL aggressively. Move every Oracle Database, Middleware, and Java workload to BYOL.
7. Lock the support uplift. Hold the support premium below three percent on the BYOL base.

Typical savings ranges

Lever	Cost line	Typical saving	Effort
Floor commit ramp	Year one and two cash	30 to 50 percent in year one	High
Term shortening	Lock in risk	Optionality reset two years earlier	Medium
BYOL conversion	License included rate	50 to 60 percent on the compute line	Medium
Escalator cap	Annual increase	2 to 4 percent per year	Low
Service inclusion	Catalog parity	Avoid the Dedicated Region step	High

What to do next

The eight step checklist is the buyer side starting position on every Cloud at Customer or Dedicated Region engagement.

1. Inventory the workloads in scope. List every workload that the on premise OCI is expected to host.
2. Sum the peak OCPU demand. Build a demand curve across the four year horizon.
3. Map service catalog requirements. Identify any service that only Dedicated Region carries.
4. Model the BYOL math. Apply the two for one core factor on x86 and the one for one factor on Arm.
5. Run a floor commit utilization model. Compare year one demand against the floor commit.
6. Cap the annual escalator. Push for three percent or below.
7. Insert a partial exit clause. Protect the half term scope reduction.
8. Document the partition posture. Confirm the OCPU level treatment and avoid VMware on the racks.

Frequently asked questions

How Redress engages on Oracle on premise cloud

Redress runs both engagements inside the Vendor Shield subscription, the Renewal Program, the Benchmark Program, and the Software Spend Assessment. Every engagement is led by a former Oracle commercial executive on the buyer side.

Read the related benchmarking, about us, locations, and contact pages.