Access Official Technical PDF

Share Summary:LinkedInEmail

For industrial organizations, moving database workloads to a public cloud is not always practical. Data residency rules, application latency, connections to on-premises systems and internal security policies can all influence infrastructure decisions. Exadata Cloud at Customer is designed to bring Oracle Exadata database services into a customer data center while using cloud-style management and consumption models. For automation and operations teams, its value depends on understanding where management responsibilities sit, how systems connect, and what scaling and lifecycle operations involve.

What Exadata Cloud at Customer changes

Traditional on-premises infrastructure is typically purchased and managed by the customer. In the model described for Gen 2 Exadata Cloud at Customer, Oracle supplies and manages the Exadata infrastructure, while the equipment is deployed in the customer’s data center. The customer operates database virtual machines and databases, using Oracle-provided automation for supported lifecycle tasks.

This arrangement is intended for organizations that want database-as-a-service operations without relocating the database platform to a public cloud data center. It can be relevant where databases support manufacturing applications, enterprise resource planning, analytics, or other systems closely connected to local application servers. The source material describes Oracle Database compatibility as supporting existing applications that use Oracle Database.

The service combines a hardware and software platform with cloud interfaces, APIs, and a subscription approach. It is therefore neither a conventional customer-managed database machine nor simply a public-cloud database service: infrastructure location, administration, and commercial model are combined differently.

Exadata Cloud at Customer architecture and responsibilities

The Gen 2 design places database servers, storage servers, and the internal high-speed fabric in the customer facility. Database workloads run in customer-managed virtual machines, referred to as DomU in the source material. Oracle manages the underlying infrastructure, including the hypervisor, servers, storage, and internal networking. Customers retain access to their database VMs and database administration credentials; Oracle operations staff do not access the customer VMs or databases under the described management model.

A key architectural component is the control plane. The public-cloud control plane provides the interface for database and virtual-machine lifecycle operations, while lightweight local control plane servers in the rack communicate with it through a secure tunnel. The connection is initiated outbound from the local servers and uses port 443, a design intended to simplify firewall configuration. Those servers relay commands to agents in customer VMs and also support monitoring and telemetry.

According to the overview, certain operations—including scaling, backup, and restore—can continue if the WAN connection to Oracle is unavailable. This does not mean every management function is independent of that connection; the architecture uses the public-cloud control plane for lifecycle requests.

Lifecycle automation and operational boundaries

Administrators can invoke supported automation through a web interface, REST APIs, command-line tools, SDKs, or Terraform. These options allow teams to perform one-off tasks through a console or incorporate operations into scripts and infrastructure-as-code workflows.

Documented automated tasks include creating and deleting VM clusters, adjusting OCPU allocations, backing up VMs, and creating, starting, stopping, backing up, restoring, or patching databases and Grid Infrastructure. The overview also describes patch rollback and encryption-related controls. The exact task set depends on the service implementation and software version; the 2019 document also distinguishes operations performed through cloud interfaces from functions invoked within a VM.

Responsibility is divided rather than transferred wholesale. Oracle owns and maintains the infrastructure layer, while customers manage database configuration, schemas, users, access, and encryption keys. Customers can also install and manage additional software in their VMs. Operations teams should define this boundary clearly in runbooks, change controls, incident procedures, and access policies.

Connecting industrial applications and scaling capacity

Database servers connect to the customer network so application servers can communicate with databases using the organization’s established network environment. The overview describes direct connections through customer switches, support for flexible VLAN configurations, and the option to separate client and backup traffic. This can help align the database platform with existing application connectivity, although network design, address allocation, routing, and security remain important deployment tasks.

The service separates the deployed infrastructure footprint from the active OCPU capacity used by database software and services. Customers can increase or reduce enabled OCPUs online, including through APIs or scheduled and programmatic workflows. The overview describes hourly OCPU adjustments and scaling without downtime. This model may help organizations align enabled capacity with changing demand, such as scheduled workloads or seasonal peaks, but it does not remove the need to size the underlying system for anticipated requirements.

Commercial options described include subscribing to database software and service or bringing eligible existing database licenses and subscribing to the service. Infrastructure is covered by a separate subscription. Because licensing and commercial terms affect total cost, teams should evaluate them alongside capacity plans and expected usage rather than treating elastic scaling alone as a cost guarantee.

Deployment planning and practical considerations

Deployment requires customer-provided data-center space, power, network connections, network configuration details, and IP addresses. The described installation process includes connecting the system to the customer network, establishing control-plane connectivity, associating the system with the cloud account, and validating the environment before handover.

Database migration can use logical approaches such as Data Pump or GoldenGate, or physical approaches such as RMAN, transportable technologies, and Data Guard. The appropriate path depends on the existing database, migration requirements, and operational constraints. Backup options in the overview include local Exadata storage, local NFS, on-premises recovery infrastructure, and Oracle public-cloud object storage.

Conclusion: align the platform with operating requirements

Exadata Cloud at Customer offers a way to run Oracle database services in a customer data center while adopting cloud-oriented lifecycle tools and a usage-based model for active database capacity. Its practical fit depends on more than database performance: organizations must assess network connectivity, control-plane access, security boundaries, migration methods, backup design, licensing, and the division of operational responsibilities. For industrial automation environments, mapping those considerations to application dependencies and plant or enterprise change processes is essential before deployment.

Access Official Technical PDF

About The Author