We’re sorry. We could not find a match for your search.

We suggest you try the following to help find what you're looking for:

  • Check the spelling of your keyword search.
  • Use synonyms for the keyword you typed, for example, try “application” instead of “software.”
  • Start a new search.
Contact Us Sign in to Oracle Cloud

Hybrid Cloud

Oracle Hybrid Cloud Offers Location, Choice, and Control

Hybrid cloud and edge computing are transforming and expanding usage of the cloud. Requirements for data sovereignty, security, latency, and field deployability have prevented adoption of public cloud for many mission-critical applications. Oracle Cloud’s hybrid and edge offerings address customer requirements for specialized deployment, disconnected and intermittently-connected operation, low latency and high performance, as well as data locality and security.

All products require tradeoffs, but to date, cloud offerings have been unnecessarily constrained. An approach focusing on large hyperscale cloud regions providing a large number of services required great expense to build, with very limited locations. Smaller scale provided greater flexibility of location, but often required tethering to the public cloud to operate. Customers with the "wrong" mix of needs around data sovereignty, low-latency, broad use cases, or level of control were left on the sidelines.

Hybrid Cloud Diagram

At Oracle Cloud, we made a different set of choices to serve our customers’ requirements for location flexibility and control, while not sacrificing performance, security, and value. Our portfolio of cloud, hybrid cloud, and edge offerings are helping customers adopt the cloud at their own pace and address hyperlocal data sovereignty rules and stringent IT SLAs. While our commercial and government public cloud regions address many customers and use cases, an increasing array of governments and organizations wanted our advanced cloud services on-premises or in remote locations.

Oracle Dedicated Region Cloud@Customer

Dedicated Region Diagram

Hyperscale public cloud regions are the standard platform for providing cloud services with the efficiencies of scale. Large public regions are powerful, but they have been, by necessity, limited in number because of their size and expense. These large data centers service an entire global region, in many cases across national boundaries. Applications that require data sovereignty, low-latency communication and processing, and specific security needs may not be well served by traditional hyperscale public cloud regions.

Oracle Cloud Infrastructure (OCI) has made technical and business choices that empower our customers to address these requirements.

Fully Managed Cloud Region

Oracle Dedicated Region is a fully managed cloud region that brings all of our cloud services into customers’ data centers in a self-contained model. Customers get the exact same architecture, billing models, operations, security, and services as in the public regions of Oracle Cloud Infrastructure (including Oracle’s enterprise cloud applications). No other cloud provider even comes close to delivering such a broad set of cloud services in a customer’s data center today.

In contrast, AWS Outposts and Azure Stack Hub have a small subset of services, tethered to their public clouds, as shown in the table below.

Oracle Dedicated Region

AWS Outposts[1]

Microsoft Azure Stack Hub[2]

# of Services Over 65 services. ALL services offered in Oracle’s public cloud regions. 17 services. Small subset of public cloud. 15 services. Small subset of public cloud.
  • Same as Oracle Public Cloud
  • Oracle is responsible for managing the end-to-end hardware, IaaS, and PaaS services (break-fix, patching, and upgrades, etc.)
  • Same as AWS Public Cloud
  • AWS is responsible for managing the end-to-end hardware, IaaS, and PaaS services (break-fix, patching and upgrades, etc.)
  • Different than Azure Public Cloud
  • Customer’s Azure Stack Hub Operators are responsible for managing the IaaS & PaaS services (break-fix, patching and upgrades)
  • HW vendors are responsible for the HW components
Availability SLA 99.95% includes planned downtime N/A N/A
Performance SLA Covered (Disk IOPS and Network) N/A N/A
Manageability SLA Covered (API Error Rate) N/A N/A
Connectivity Fully self-contained region; metadata and API operations remain local to data center. Upon loss of connectivity API availability is not degraded. Simple FastConnect connection for LAN access to region. Metadata and control-plane operations not local to data center. Upon loss of connectivity API availability is degraded. Setup two link aggregation groups (LAG) and VLANs on a per rack basis for LAN access. Disconnected with feature trade-offs. Complex border, switch, DNS and firewall setup for LAN/outbound access.
Data Sovereignty
  • Customer data always stay on-premises
  • Customer’s onsite data is backed up to another of the customer’s own data centers
  • Must be connected to a supported AWS Region
  • Instance IDs, monitoring metrics, metering records, tags, etc. sent to the AWS Region
  • “EBS snapshots of EBS Volumes on Outposts are stored by default on Amazon S3 in the Region”
  • Database metrics are stored in AWS CloudWatch
  • In Disconnected Mode, no data sent to Azure public cloud, but many features are impaired or unavailable
Economics Pay for what you use: purely OpEx. Oracle manages capacity at no additional cost, as long as customers meet their spend commitment. Pay by the rack. Capacity planning customers’ responsibility. Pay for Azure Stack Hub services, and separately for OEM hardware. Pay for installation, upgrade, and decommission.
Compliance SOC 1, SOC 2, ISO 27001, ISO 27017, and implements controls set in National Institute of Standards and Technology (NIST) 800-53 rev4 publication N/A N/A
Support 24/7 remote support part of the consumption-based commitment. Requires separate purchase of “Enterprise Support” for 24/7 remote support. Contact Azure Stack Hub operator for support
NRI logo
We have finished deploying Oracle Dedicated Region Cloud@Customer in our data center, and will next migrate our SaaS applications for the financial industry from on-premises to Oracle Dedicated Region Cloud@Customer. With Oracle Dedicated Region Cloud@Customer, we’ve been able to achieve the same level or even better performance than our existing on-premises system while continuing to maintain the high level of availability we need. We look forward to launching our services to customers, as well as our ongoing relationship with Oracle to operate and maintain this Dedicated Region and continuing to drive our digital transformation together.

Tomoshiro Takemoto Senior Corporate Managing Director, Nomura Research Institute

OCI public and government cloud regions

OCI’s architectural choices have allowed us to rapidly grow a network of over 30 cloud regions, including multiple regions with Azure Interconnect for multicloud workloads. Customers around the world are running their businesses on our public cloud regions with higher performance, savings from globally consistent low pricing, higher disaster protection from multiple distributed regions in many countries, and greater data sovereignty.

Global Cloud Data Center Infrastructure map
France 2
Milan, Italy
Israel 2
Chile 2
Saudi Arabia 2
South Africa Central (Johannesburg)
Commercial Sweden (Stockholm)
Commercial Singapore (Singapore)
Commercial France South (Marseille)
Commercial Israel Central (Jerusalem)
Commercial US West (San Jose) OCI–Azure Interconnect
Commercial US West (Phoenix)
Government US Gov West (Phoenix) DISA Impact Level 5 Authorization
Government US DoD West (Phoenix) FedRAMP High Joint Authorization Board
Commercial Canada Southeast (Toronto) OCI-Azure Interconnect
Government US DoD North (Chicago) DISA Impact Level 5 Authorization
Commercial US East (Ashburn) OCI-Azure Interconnect
Government US Gov East (Ashburn) FedRAMP High Joint Authorization Board
Government US DoD East (Ashburn) DISA Impact Level 5 Authorization
Commercial Canada Southeast (Montreal)
Commercial Chile (Santiago)
Commercial Brazil Southeast (Vinhedo) OCI-Azure Interconnect
Commercial Brazil East (Sao Paulo)
Commercial UK West (Newport)
Government UK Gov West (Newport)
Commercial UK South (London) OCI-Azure Interconnect
Government UK Gov South (London)
Commercial Netherlands Northwest (Amsterdam) OCI-Azure Interconnect
Commercial Germany Central (Frankfurt) OCI-Azure Interconnect
Commercial Switzerland North (Zurich)
Commercial Saudi Arabia West (Jeddah)
Commercial UAE East (Dubai)
Commercial UAE Central (Abu Dhabi)
Commercial India West (Mumbai)
Commercial India South (Hyderabad)
Commercial South Korea Central (Seoul)
Commercial South Korea North (Chuncheon)
Commercial Japan East (Tokyo) OCI-Azure Interconnect
Commercial Japan Central (Osaka)
Commercial Australia East (Sydney)
Commercial Australia Southeast (Melbourne)

  • Australia
  • Two regions (Sydney, Melbourne)
  • Brazil
  • Two regions (Sao Paulo, Vinhedo)
  • Canada
  • Two regions (Toronto, Montreal)
  • Chile
  • One region (Santiago)
  • European Union
  • Four regions (Amsterdam, Frankfurt, Marseille, Zurich)
  • India
  • Two regions (Mumbai, Hyderabad)
  • Israel
  • One region (Jerusalem)
  • Japan
  • Two regions (Tokyo, Osaka)
  • Saudi Arabia
  • One region (Jeddah)
  • Singapore
  • One region (Singapore)
  • South Africa
  • One region (Johannesburg)
  • South Korea
  • Two regions (Seoul, Chuncheon)
  • Sweden
  • One region (Stockholm)
  • United Arab Emirates
  • Two regions (Dubai, Abu Dhabi)
  • United Kingdom
  • Two regions (London, Newport)
  • United States
  • Three regions (Ashburn, Phoenix, San Jose)
Commercial planned

  • Milan, Italy
  • Chile 2
  • Colombia
  • France 2
  • Israel 2
  • Mexico
  • Saudi Arabia 2
  • Spain

  • United Kingdom
  • Two U.K. Gov regions (London, Newport)
  • United States
  • Two U.S. Gov regions (Ashburn, Phoenix) and three U.S. Department of Defense regions (Ashburn, Chicago, Phoenix)
When we architected Oracle Cloud Infrastructure, the tens of thousands of enterprise customers that Oracle serves across the globe made it clear that they wanted public cloud services for their most demanding enterprise workloads in their country or geo-area. Data residence requirements, country-specific compliance requirements, and network latency were some of the key driving forces. To thrill our customers, our strategy has been to launch right-sized regions closer to where our customers want them and scale them dynamically while preserving public cloud characteristics, such as elasticity, high reliability, and operational excellence. To serve the needs of our customers, we had to launch regions much faster than other cloud providers.

Pradeep Vincent Vice President and Architect, Oracle Cloud Infrastructure

Government Cloud Regions

Oracle’s US government cloud includes two FedRAMP regions for US federal, state, and local government use; three regions dedicated specifically for the US Department of Defense (DoD) and federal customers who have workloads subject to DISA SRG impact level guidance; and National Security Regions especially designed and built for US Defense and Intelligence communities with associated accreditations (DoD Impact Level 6, ICD 503). Oracle also provides a dual-region sovereign cloud for the UK government and public sector.

Oracle Cloud VMware Solution

Cloud VMware Solution Diagram

VMware is the foundation of the majority of enterprise data centers. To date, enterprises have chosen to either deploy VMware with full control on-premises or to delegate control to an MSP or hosting provider. VMware services in the cloud provided benefits not found in traditional infrastructure environments, including just-in-time provisioning, elastic capacity scaling both up and down, and pay-for-usage models. However, traditional cloud services required relinquishing full control over administration, access, and upgrades.

OCI has delivered a cloud VMware solution that offers these benefits while still giving VMware administrators full control over their environment.

VMWare Solution Diagram

Oracle Cloud VMware Solution (OCVS) is the only solution on the market that provides a native VMware-based cloud environment that offers complete control using familiar VMware tools.

VMware workloads to the cloud without having to modify them. They also want to gain scale and agility while maintaining continuity with existing VMware-based tools, processes, and policies. Customers can use Oracle VMware Solution to move or extend on-premises VMware-based workloads to OCI across all regions, including our on-premises Dedicated Regions, without rearchitecting applications or retooling operations.

OCVS offers greater customer control of the VMware environment, with the elasticity of the cloud, and is available in more regions around the world:

Oracle Cloud VMware Solution

VMware Cloud on AWS[3]

Azure VMware Solution[4]

Google Cloud VMware Engine[5]

Operating Model Customer-managed and controlled Managed services by VMware and AWS Managed services by Microsoft Managed services by Google
Maximum nodes per VMware SDDC environment 64 16 16 32
Security Customer owns root credentials. Oracle doesn’t have access to root credentials or metadata. AWS retains root credentials and metadata perpetually. Azure retains root credentials and metadata perpetually. GCP retains root credentials and metadata perpetually.
Billing Consolidated Separate bills from AWS and VMware Unspecified Consolidated
Support Oracle VMware and AWS Microsoft provides support Google provides support
Updates, patches, and upgrades Customer controls when and whether to upgrade. AWS controls and decides. Azure controls and decides. GCP controls and decides.
SDDC vCenter access Full Administrator access Restricted access Restricted access Restricted access
Region Availability 22 OCI regions
2 OCI Gov regions
17 AWS regions 9 10
On-Premises Cloud Yes. Oracle Dedicated Region No. VMware Cloud on AWS Outposts (In Preview) No. Not available on Microsoft Azure Stack Hub No. Google Anthos cannot run IaaS on-premises
Entel logo
Entel migrated more than 60 production applications to Oracle Cloud Infrastructure, including our mission-critical customer engagement, business intelligence, and risk management systems. We have a strong technical team, including many VMware experts, so a managed VMware service is not helpful. We need administrative access to the VMware software so we can maintain control of every aspect of the VMware environment. Oracle Cloud VMware Solution offers my team the exact same VMware experience they had in the data center but with cloud elasticity and access to Oracle Cloud services, providing a fast, easy, and cost-effective path to migrate our VMware-based applications to the cloud. As a result, performance has increased dramatically, and downtime is close to zero.

Helder Branco Chief Technology Officer, Entel

Hybrid Cloud with Oracle Exadata Cloud@Customer

Hybrid Cloud with Exadata Diagram

A common hybrid cloud pattern is a “remotely-tethered” rack or server on the customer’s premises. The "control plane" or service management runs in the public cloud to take advantage of existing services, governance, and operational control, while hardware and data remain on-premises. This model gives tremendous operational advantages, but choices must be made where data, metadata, and backups are located, and how interruption of connectivity affects operations and control.

Exadata Cloud@Customer at Data Center

Oracle Exadata Cloud@Customer delivers high-performance Exadata Cloud Service capabilities directly to a customer’s data center. The service helps address requirements for strict data sovereignty and security by retaining data securely on-premises while shifting operational management to an OCI-managed cloud service. Once provisioned, databases remain fully operational even when disconnected from the cloud. Optional Oracle Autonomous Database capabilities eliminate up to 80% of manual database operations. The solution is delivered with an all-inclusive or bring-your-own license subscription model, and the ability to scale up or down seamlessly as needed.

Oracle Exadata C@C RDS on AWS Outposts[6] Azure Stack Hub[7]
Min SQL read latency 19 µs ≥ 1000 µs OEM variable
Max SQL read IOPS 12 million 960,000 OEM variable
Max SQL throughput 300 GB/s 28.5 GB/s OEM variable
Max database size 476 TB 64 TB 4.75 TB
Time in production 3.5 years < 1 year 3.5 years
Purpose-built for database workloads Database-aware Flash Cache and persistent memory, offload SQL to storage; Hybrid Columnar Compression Non-optimized infrastructure for database workloads Non-optimized infrastructure for database workloads
Enterprise database capabilities Oracle RAC, Oracle Active Data Guard, and advanced database security; Oracle Multitenant, multiworkload, multidata types in one engine No AWS native DBs (Aurora, Redshift, etc.); RDS only for open source MySQL, PostgreSQL No DBaaS on Azure Stack Hub
Autonomous Database: self-driving, self-securing, self-repairing
No downtime for maintenance
Delivered and supported as a service by cloud provider
Service Level Objective 99.95%, including planned downtime
Samsung logo
Back in 2010, we adopted the first Exadata platform to improve a display manufacturing system. Now 10 years later, we have implemented nearly 300 Exadata systems for our customers in manufacturing, financial services, construction and engineering and public and private sector services. Aligning with our digital innovation strategy and our journey to enterprise cloud, we have now adopted the first Exadata Cloud@Customer in one of our datacenters and look forward to deploying Autonomous Database.

Dr. WP Hong CEO, Samsung SDS

Oracle Roving Edge Infrastructure

Roving Edge Infrastructure Diagram

Edge computing applications reside at or beyond the boundary of a cloud region or network. Self-contained, independent operation with isolated or intermittent connectivity is key to satisfy the requirements of these remote deployments. Physical instances must have enough processing, throughput, and storage to handle intensive workloads in limited timeframes.

Roving Edge Infrastructure on Ruggedized Device

Oracle Roving Edge Infrastructure is a service that extends fundamental Oracle Cloud Infrastructure capabilities to a ruggedized device. This allows applications with requirements for low-latency, time-sensitivity, and high throughput to function even in locations without network connectivity.

The key component of Oracle Roving Edge is the Roving Edge Device (RED)—a high-powered and portable server node. Customers configure a RED in their OCI console with the same services, applications, and data they use in the cloud. OCI then ships the device to the customer. Devices are deployed in the field and can operate either without a network connection or connected to the internet or a local network. After the device usage is complete, the customer returns the RED to an OCI Transfer Station, where data can be synchronized with the cloud and the device is prepared to be used again. Roving Edge Devices can also be deployed in larger clusters for more resilient, longer-lived compute resources or mobile data centers.

Oracle Roving Edge Infrastructure

AWS Snowball Edge Compute Optimized[8]

Azure Stack Edge Pro[9]

CPU cores 40 26 20
DRAM 512 GB 208 GB 128 GB
Storage 61 TB NVMe
2 x 1 TB NVMe (System OS)
7.68 TB NVMe for VMs
12.8 TB NVMe for local cache
AI/ML Support NVIDIA Tesla T4 GPU NVIDIA Tesla V100 Intel Arria 10 GX FPGA or NVIDIA T4 GPU
Clustering 5-15 node clusters 5-10 node clusters  

OCI services for hybrid cloud architecture

Learn about OCI’s offerings for on-premises cloud environments that support hybrid cloud architecture.

Understanding hybrid cloud deployments

Hybrid cloud architecture uses both a public cloud provider, such as OCI, and on-premises infrastructure to meet enterprise IT needs. This approach provides the flexibility to move some workloads to the public cloud while retaining others on-premises.

According to industry analyst IDC, “hybrid cloud has become central to successful digital transformation efforts.” This architecture is gaining importance as certain workloads cannot be moved to public cloud data centers for reasons that include data sovereignty, data residency, network latency, and infrastructure performance. Oracle Cloud@Customer brings the complete portfolio of cloud services into customer data centers, enabling hybrid cloud architectures.


Many large-scale data center users have been working through plans for cloud migration, but struggle to move gradually, keeping processing on-premises and moving applications and data in a stepwise manner to the cloud. Exadata Cloud@Customer and Oracle Dedicated Region Cloud@Customer overcome this issue by, in effect, bringing the cloud to the customer.

Carl Olofson
Research Vice President, IDC


Complete cloud capabilities within your data center

Oracle Dedicated Region is a completely managed cloud region that brings all of Oracle’s second-generation public cloud services and Oracle Fusion SaaS applications into your data center, and runs these without data sovereignty concerns. The solution enables autonomous operations that eliminate human error and maximize the security of your data.

Oracle Dedicated Region can be used with OCI’s public cloud services for hybrid cloud deployments and can be interconnected with other cloud providers. With Oracle Dedicated Region, customers increase agility and innovation by accessing new cloud services on-premises as soon as Oracle makes them available in public cloud regions—all while lowering costs with OCI’s leading price-performance. Further, you can leverage the exact same SLAs, APIs, and tools available in Oracle’s public cloud, and experience low-latency interaction with your on-premises applications and databases.

Advanced machine-learning–based database for on-premises deployments

Oracle Autonomous Database on Exadata Cloud@Customer combines all the benefits of having Oracle Exadata in your data center with the simplicity of an autonomous cloud service. It helps you meet strict data sovereignty and security requirements and eliminate many manual database and infrastructure management tasks, while supporting on-premises enterprise applications with high Exadata performance.

The solution can be used with OCI’s public cloud services for hybrid cloud deployments and can be interconnected with other cloud providers. Oracle Autonomous Database enables you to consolidate databases with autonomous operations, and to build a modern enterprise data warehouse without the challenge and cost of moving your on-premises data. IT teams benefit from machine-learning–based automation of provisioning, configuring, and tuning—reducing management cost and eliminating downtime with automated error detection, correction, and patching.

High-performance managed database for your data center

Oracle Exadata Cloud@Customer combines the performance of Oracle Exadata with the simplicity, flexibility, and affordability of a managed database service in your data centers. It’s the simplest way to move your existing Oracle databases to the cloud, providing full compatibility with your existing Exadata systems and Oracle Exadata Cloud Service.

The solution can be used with OCI’s public cloud services for hybrid cloud deployments and can be interconnected with other cloud providers. Exadata Cloud@Customer enables customers to meet data sovereignty and low-latency requirements in existing data center resources while achieving higher database consolidation and scaling that isn’t possible on-premises.

Cloud computing and storage services at the edge

Oracle Roving Edge Infrastructure accelerates deployment of cloud workloads outside the data center. Ruggedized Oracle Roving Edge Devices (Oracle REDs) deliver cloud computing and storage services at the edge of networks and in disconnected locations, allowing faster processing close to the data source and enabling faster insights into the data.

Existing Oracle Cloud Infrastructure (OCI) virtual machines (VMs) and object storage can be synchronized to Oracle Roving Edge devices using the same portal and tenancy tooling as our public regions. Roving Edge Devices can be used for fast data collection and processing in the field, application deployment to remote facilities, development/test in cloud and deployment to edge, and artificial intelligence/machine learning (AI/ML) at the edge.


Adopt hybrid cloud with OCI

Browse reference architectures and use cases, validate technical requirements, and learn best practices on hybrid cloud architectures.

Migrate to the Cloud
Build a hybrid software-defined data center (SDDC) by extending on-premises VMware to OCI
Maximize Database Insights
Deploy a hybrid disaster recovery topology for on-premises Oracle Exadata
PeopleSoft deployment on OCI
Deploy a hybrid disaster recovery topology for an on-premises database
Setup Hyperion
Deploy an Oracle API Gateway service in a hybrid environment

Start a free Oracle Cloud Infrastructure (OCI) trial.

As of February 2021:

[1] https://aws.amazon.com/outposts/features/, https://aws.amazon.com/outposts/faqs/, https://aws.amazon.com/outposts/pricing/

[2] https://azure.microsoft.com/en-us/products/azure-stack/hub/#faqs, https://azure.microsoft.com/en-us/global-infrastructure/services/?products=all®ions=azure-stack-hub, https://docs.microsoft.com/en-us/azure-stack/operator/azure-stack-disconnected-deployment?view=azs-2002, https://docs.microsoft.com/en-us/azure-stack/operator/azure-stack-network?view=azs-2002, https://docs.microsoft.com/en-us/azure-stack/operator/azure-stack-billing-and-chargeback?view=azs-2002

[3] https://configmax.vmware.com/guest?vmwareproduct=VMware%20Cloud%20on%20AWS&release=VMware%20Cloud%20on%20AWS&categories=68-0,52-0,3-0,53-0,54-0,55-0,56-0,57-0,58-0,75-0,76-0, https://www.vmware.com/content/dam/digitalmarketing/vmware/en/pdf/partners/vmware-cloudtm-on-aws-technical-overview.pdf, https://www.vmware.com/content/dam/digitalmarketing/vmware/en/pdf/support/vmw-cloud-aws-service-description.pdf, https://cloud.vmware.com/vmc-aws/faq, https://docs.vmware.com/en/VMware-Cloud-on-AWS/services/com.vmware.vmc-aws-operations/GUID-7458E592-FC6E-43A8-B0BE-505B1435D12D.html, https://aws.amazon.com/vmware/faqs/

[4] https://docs.microsoft.com/en-us/azure/azure-vmware/faq, https://azure.microsoft.com/en-us/global-infrastructure/services/?regions=all&products=azure-vmware

[5] https://cloud.google.com/vmware-engine/docs/concepts-private-cloud, https://cloud.google.com/vmware-engine/docs/concepts-permission-model, https://cloud.google.com/vmware-engine/docs/concepts-vmware-components, https://cloud.google.com/blog/topics/hybrid-cloud/announcing-google-cloud-vmware-engine, https://cloud.google.com/vmware-engine/docs/concepts-permission-model, https://cloud.google.com/vmware-engine/

[6] https://aws.amazon.com/outposts/features/, https://aws.amazon.com/outposts/faqs/, https://aws.amazon.com/rds/faqs/, https://aws.amazon.com/outposts/pricing/, Performance calculated for maximum 12 servers from https://aws.amazon.com/about-aws/whats-new/2019/04/amazon-rds-for-oracle-now-supports-64tib/, https://docs.amazonaws.cn/en_us/AmazonRDS/latest/UserGuide/CHAP_Storage.html, https://docs.aws.amazon.com/AmazonRDS/latest/UserGuide/Concepts.DBInstanceClass.html

[7] https://azure.microsoft.com/en-us/products/azure-stack/hub/, https://docs.microsoft.com/en-us/azure-stack/user/azure-stack-acs-differences?view=azs-2002

[8] https://docs.aws.amazon.com/snowball/latest/developer-guide/specifications.html#specs-compute-optimized and https://docs.aws.amazon.com/snowball/latest/developer-guide/UsingCluster.html as of January 2021

[9] https://docs.microsoft.com/en-us/azure/databox-online/azure-stack-edge-gpu-overview, https://azure.microsoft.com/en-us/pricing/details/azure-stack/edge/#azureStackEdgePro, and https://docs.microsoft.com/en-us/azure/databox-online/azure-stack-edge-gpu-technical-specifications-compliance as of January 2021