Compute Cloud@Customer FAQ

General

What is Oracle Compute Cloud@Customer?

Oracle Compute Cloud@Customer is a fully managed, rack-scale distributed cloud platform that lets customers use Oracle Cloud Infrastructure (OCI) Compute and GPU capability in a location of their choice. Remotely managed by Oracle, Compute Cloud@Customer lets customers gain the benefits of cloud automation and economics while addressing data residency requirements by controlling where their data is stored and processed.

What are the benefits of running workloads on Oracle Compute Cloud@Customer?

With Oracle Compute Cloud@Customer, customers can

• Develop applications once and run them anywhere in Oracle’s distributed cloud without changes using consistent services, APIs, SDKs, and infrastructure-as-code capabilities
• Deploy fully managed cloud infrastructure at scale in distributed data centers and dynamically consume OCI Compute and storage resources for Oracle, third-party, and custom applications while only paying consumption costs for the resources they use
• Modernize the infrastructure for cloud native and traditional applications by running application and middleware tiers on Compute Cloud@Customer and the Oracle Database tier on Oracle Exadata Cloud@Customer
• Address data residency and other regulatory requirements by controlling the location and security of data and its processing

What configurations are available for Oracle Compute Cloud@Customer?

What OCI Compute services are available on Oracle Compute Cloud@Customer?

Oracle Compute Cloud@Customer supports OCI Compute with flexible virtual machine (VM) shapes and GPU shapes. Each VM supports 1 to 96 OCPUs and 1 GB to 64 GB of memory per OCPU, with a maximum of 960 GB of memory per VM. For more details on supported guest operating systems, please refer to the product documentation.

What OCI storage services are available on Oracle Compute Cloud@Customer?

Oracle Compute Cloud@Customer supports block, object, and file storage services on high-capacity (balanced) storage trays and block storage on high performance storage trays.

What OCI networking services are available on Oracle Compute Cloud@Customer?

Oracle Compute Cloud@Customer supports the same networking constructs as OCI, including virtual cloud networks, subnets, gateways, security lists, route tables, and more.

Where can I find the list prices for Oracle Compute Cloud@Customer and the services running on it?

Noncontractual list prices with local currency translations for Oracle Compute Cloud@Customer infrastructure are available on the product page. Prices for OCI services that run on Oracle Compute Cloud@Customer are available on the OCI price list page.

For an official quotation, please contact Oracle Sales.

How do I pay for the resources I consume?

Pricing for Oracle Compute Cloud@Customer has two components: a nonmetered charge for dedicated infrastructure and consumption-based service costs. For the nonmetered dedicated infrastructure charge, customers pay based on their selected configuration for the duration of their infrastructure subscription. For consumption-based services, customers are charged OCI-equivalent pricing according to their consumption of the metered services. Compute Cloud@Customer also supports the Universal Credits model with license included and bring your own license (BYOL) pricing for metered services. For more information, please check the Oracle Compute Cloud@Customer page.

Can I connect Oracle Exadata Cloud@Customer to Compute Cloud@Customer?

Yes. Oracle Compute Cloud@Customer supports a unique direct connection to one or more Oracle Exadata Cloud@Customer or Oracle Exadata Database Machine platforms with up to 800 Gb/sec of dedicated bandwidth. By running the applications or middleware layer on Compute Cloud@Customer and the database layer on Exadata Cloud@Customer or Exadata Database Machine, customers can benefit from the lowest latency and highest performance connectivity.

How scalable is Oracle Compute Cloud@Customer?

Oracle Compute Cloud@Customer is highly scalable, with compute scaling from 552 OCPUs to 6,072 OCPUs and storage scaling from 150 TB to 3.65 PB. Both compute and storage can be scaled independently of each other.

What are some key capabilities of Oracle Compute Cloud@Customer?

• Oracle owns, manages, and operates Compute Cloud@Customer infrastructure, while customers manage guest operating systems and consume common OCI services using consumption-based economics.

• Compute Cloud@Customer brings a cloud experience to customer data centers. Each platform is deployed behind the customer’s firewall, enabling them to address residency requirements for their data and its processing.

• Compute Cloud@Customer can be deployed in existing customer data centers to minimize latency when connecting cloud compute resources to other data and processing resources in those data centers.
• Compute Cloud@Customer can be directly connected to an Exadata Cloud@Customer or Exadata Database Machine in the same data center, minimizing the latency applications experience when accessing Oracle Database services running on Exadata.
• Compute Cloud@Customer can be deployed in distributed data centers that are close to end user customer locations to support applications that require real-time responses, such as Internet of Things, gaming, and telco services.

• Customers use identical OCI APIs, SDKs, and management to develop and deploy workloads in the public cloud or on Compute Cloud@Customer.
• Customers can use Terraform scripts to deploy workloads using the OCI command-line interface, with portability across Oracle’s distributed cloud.

What GPU configuration does Oracle Compute Cloud@Customer GPU Expansion support?

Oracle Compute Cloud@Customer GPU Expansion features NVIDIA L40S GPUs. It lets customers bring high performance AI, machine learning (ML), and database capabilities into their own data centers, all with the control and security of an on-premises environment.

How scalable is the GPU expansion for Oracle Compute Cloud@Customer GPU Expansion?

Oracle Compute Cloud@Customer GPU Expansion can scale from 4 to 48 NVIDIA L40S GPUs.

What are the key benefits of customers running GPUs on Oracle Compute Cloud@Customer?

• Multiple NVIDIA L40S GPUs coupled with scalable compute and storage capabilities provide high performance for demanding workloads across AI/ML, 3D graphics, 4K streaming, and high performance computing (HPC).

• Train in the cloud and deploy to Compute Cloud@Customer, enabling a unified workflow from edge to cloud. Flexible deployment options in customer data centers address needs for data residency with a hybrid cloud solution.

• Scalable high performance supports large language model (LLM) training and fine-tuning, generative AI inferencing, deep learning, and computer vision tasks, with support for the latest libraries and tools to optimize performance.

• Enterprise-grade infrastructure runs mission-critical applications with high uptime and reliability. The scalable design of Compute Cloud@Customer supports data-intensive workloads that can scale from terabytes to petabytes as your business needs grow.

What are some targeted workloads for Oracle Compute Cloud@Customer GPU Expansion?

• Generative AI inferencing: Real-time inferencing for multimodal generative AI pipelines (text, images, audio, and video)
• LLM training and fine-tuning: Accelerated performance for fine-tuning medium LLMs and training small LLMs with NVIDIA Transformer Engine and FP8 support
• Graphics-intensive and virtual desktop infrastructure applications: 3D graphics and rendering workflows with NVIDIA RTX and ray tracing capabilities
• Digital twins using NVIDIA Omniverse: Development and operations for complex 3D industrial digitalization workflows
• Media streaming: Increased encode/decode density and AV1 support for 4K video streaming
• HPC: Scientific data analysis and simulation workloads with FP32 support

Where can I find technical information for Oracle Compute Cloud@Customer?

Please refer to the datasheet (PDF) to understand more about Oracle Compute Cloud@Customer.