Oracle Corp. last week made some cloud announcements that represent the most ambitious attempt yet, among a number of others, to bring a full cloud experience to enterprises’ on-premises data centers.

The most noteworthy of the announcements were the availability of Autonomous Database on Exadata Cloud@Customer (including Autonomous Data Guard) and Oracle Dedicated Region Cloud@Customer. Together, they represent a radical approach to bringing cloud services on-premises. Our initial take is that customers running mission-critical Oracle applications can cut costs by up to 40% with these offerings.

Oracle Cloud@Customer announcements

On July 8, Oracle made two strategic announcements that span its cloud portfolio:

Autonomous Database on Exadata Cloud@Customer

Oracle’s advanced Autonomous Database is now available on-premises. This new offering brings its machine-learning powered database to customer data centers. The starting commitment for Autonomous Database on Exadata Cloud@Customer is reduced to $10,800 per month for three years, a 20% discount.

At the same time, Oracle has upgraded the infrastructure platform with the lower-latency Exadata X8M. The X8M uses persistent memory and RoCE to reduce storage latency, which is particularly important for heavy database workloads.

All the complexities of the Autonomous Database are handled over the control plane to the Oracle Cloud. No user data is shipped on the control plane.

Oracle has also announced Autonomous Data Guard, which allows seamless failover and switchover to and from a physical copy of the database. All the complexities are handled by the Autonomous Database service and orchestrated using the control plane within Oracle Cloud. This automation simplifies database operations and automates the testing of system high-availability. As a result, organizations can have increased confidence in compliance thru simplified high availability validation tests to meet internal and external auditing requirements.

Wikibon believes that Oracle Autonomous Database on Exadata Cloud@Customer will find broad acceptance from Oracle’s global 2000 customers.

Oracle Dedicated Region Cloud@Customer 

Dedicated Region Cloud@Customer makes Oracle’s entire software-as-a-service portfolio available on-premises and managed by the Oracle Cloud control plane. As with other services, no customer data is transferred across the control plane. The software supported includes Oracle’s ERP, HCM, CRM and CX packages. This provides an on-premises cloud experience for SaaS applications.

Exadata X8M hardware and software are included in the Dedicated Region. This infrastructure supports Autonomous Databases and services run on-premises and supported by Oracle over the Cloud@Customer control plane.

In addition, Oracle provides hyper-scale infrastructure on-premises, which can run bare-metal compute, VMware services, Containers for Kubernetes, and many other services.

Overall there are 50 services in total. Notably, to take advantage of this capability, customers must commit to an $18 million minimum spend over three years, or $500,000 per month. This figure may seem like “sticker shock.” However, it’s quite reasonable relative to the expense of running Oracle infrastructure with internal resources, and for large organizations, this is a relatively small investment over three years.

Oracle Cloud@Customer background

Oracle focuses on sophisticated operational software to run large-scale mission-critical applications for global 2000 organizations. The Oracle Tier-1 Database is at the heart of this software and hardware stack.

Oracle is now focusing on bringing a complete cloud experience on-premises for these sophisticated environments.

The top half of Figure 1 summarizes the Autonomous Database on Exadata Cloud@Customer, which supports Autonomous Database on-premises deployed on Exadata X8M. The announcement of Autonomous Data Guard is an important extension.

The bottom half of Figure 1 summarizes Oracle Dedicated Region Cloud@Customer, which supports Autonomous Database on-premises deployed on Exadata X8M. It also supports all 50 0f the Oracle@Cloud services on other Oracle hyperconverged infrastructure, including Oracle SaaS products running on the Oracle Cloud.

The Autonomous Database services include all the layers of the stack. These include the operating system, Exadata X8M, and the Oracle Database. The Autonomous Database is supported on both Exadata Cloud@Customer and Oracle Dedicated Region Cloud@Customer.

• Autonomous Operational Management is assisted by the Oracle Cloud control plane.
• Autonomous Database services include:
  • Autonomous Linux;
  • Oracle Autonomous Transaction Processing;
  • Oracle Autonomous Data Warehouse;
  • Autonomous Data Guard.
• In addition to the X8M architectural upgrades to Exadata Cloud@Customer, two new additions include:
  • Multiple VM Clusters per Exadata Rack, which enables organizations to share an Exadata system for production, DR and dev/test. This capability provides isolation across departments and different use cases.
  • Support for the Payment Card Industry Data Security Standard requirements (PCI-DSS Certification).

Wikibon expects other database services such as blockchain to become autonomous over time.

Automated backup is part of the Oracle Autonomous Database on Exadata Cloud@Customer and Dedicated Region Cloud@Customer. Organizations can choose where to store backups on-premises or in a cloud.

Oracle Cloud@Customer strategy

Oracle’s strategy is to migrate most customers to use the Oracle Cloud services, either on-premises or in the cloud. It can then use the volume of customers to accelerate the autonomous services provided.

Oracle Dedicated Region Cloud@Customer can run any Oracle SaaS Service such as Oracle E-Business Suite and the Oracle Cloud VMware service. This flexibility allows, for example, an in-house E-Business Suite service, which is tightly integrated with on-premises software and workflows, to be outsourced to the Oracle Cloud and retain the on-premises latency and location. Similarly, an Oracle Cloud VMware service can be run on-premises, and the provisioning and management of this service can be outsourced to the Oracle cloud.

This level of automation of all aspects of running and optimizing the stacks reduces the high-skilled staff required to operate organization IT support drastically. The business case for Global 2000 organizations running Autonomous Database on Exadata Cloud@Customer and Oracle Dedicated Region Cloud@Customer is compelling.

Comparison with AWS Outposts, Microsoft Azure Stack, Google Anthos and SaaS vendors

Amazon Web Services Inc.’s Outposts, Microsoft Corp.’s Azure Stack Hub and Google Cloud’s Anthos solutions all bring the cloud hardware and control-plane to the organization data center. These solutions all enable migrating applications running in the cloud on-premises. These solutions can provide improved latency, data sovereignty compliance and better workflow integration with other on-premises applications.

However, for sophisticated mission-critical operational software, the organization is responsible for the implementation, integration, security, updating and service level agreements of the Tier-1 database environment.

At this time, most SaaS vendors have no on-premises cloud offerings at this time. Wikibon believes that SaaS vendors will need to provide the ability to offer an on-premises version. This approach protects the organization’s data and makes it easier to integrate with other on-premises workflows. Wikibon believes that Oracle will benefit significantly from the ability to offer its SaaS portfolio in Oracle cloud and in Oracle Cloud@Customer.

AWS Outposts

AWS, for example, in its public cloud, claims that a combination of its Aurora transactional database together with functions from the PaaS layer can achieve high levels of recoverability. However, Wikibon would emphasize that the organization is responsible for designing, implementing, updating and maintaining this environment. It is not a managed AWS service.

In addition, Wikibon’s understanding is that neither AWS Aurora nor Redshift is currently available on AWS Outposts. MySQL and PostgreSQL are the only transaction and data warehouse databases currently available. In Wikibon’s opinion, none of the previously mentioned AWS databases are suitable to run large-scale mission-critical business applications with aggressive SLAs for availability, recovery and security.

Wikibon would also point out that AWS requires the back-up of all data residing on Outposts to an AWS public cloud region. It does not allow any on-premises data backup to a local storage device. This may void data residency and data sovereignty requirements of such a service.

Microsoft Azure Stack 

Microsoft starts in a better position than AWS by having a Tier-1 database, Microsoft SQL Server, which they offer on Microsoft Azure. However, the Microsoft SQL for Azure Stack Hub is a subset of the original, with most of the Tier-1 capabilities unavailable.

In addition, an organization running Oracle would need to convert the Oracle applications to run on SQL Server. Microsoft advocates this strategy but does not offer to perform the conversion service. As Wikibon has stated before, customers should avoid converting from large-scale Oracle Databases to SQL Server like the plague.

Google Anthos

Google Anthos is for completely different workloads, such as Kubernetes. These are important workloads. However, at this moment, Google does not claim that Google Anthos is a suitable platform for large-scale mission-critical systems of record. Therefore, it is not a direct competitor to Oracle Cloud@Customer offerings.

SaaS vendors

At this time, most SaaS vendors, such as Workday Inc. and Infor, have no on-premises cloud offerings at this time. Wikibon believes that SaaS vendors will need to provide the ability to offer an on-premises version. This approach protects an organization’s data and makes it easier to integrate with other on-premises workflows.

Wikibon believes that there is an opportunity for Oracle to provide a nonintrusive option to SaaS ISV vendors wanting cloud and on-premises versions, and differentiate by showing how the Oracle architecture makes the ISV customer data and resource usage data safe.

Comparison conclusions

Wikibon observes that AWS Outposts, Microsoft Azure Stack, and Google Anthos are providing hyperscale technology focused on cost rather than performance. This is a good approach for the more traditional cloud services organizations that need on-premises capabilities.

However, for large-scale mission-critical Oracle Databases, latency and scale are important. For example, the IO latency on the Oracle Exadata X8M is less than 20 microseconds. A millisecond IO latency would be very fast for standard hyperscale technology. Exadata can support more than 10 million database reads, whereas other cloud on-premises providers would struggle to achieve 1 million.

The additional advantage of high performance is that the number of database licenses required to run an application is reduced. The cost of Tier-1 database licenses from any vendor is significantly higher than the infrastructure costs.

Finally, the cost of conversion from an Oracle mission-critical database to an alternative database environment is also an expensive and extremely high-risk undertaking. For example, Amazon has taken over five years so far to migrate from Oracle Databases and thousands of person-years of conversion services. Case studies of global 2000 AWS customers successfully migrating large-scale mission-critical databases to AWS or other traditional cloud vendors are conspicuous by their absence.

As Wikibon has stated before, organizations should avoid converting from Oracle Database to any cloud database at all costs.

Business case for Oracle Cloud strategy

Both the short-term and long-term business cases for migrating to Oracle Autonomous Database on Exadata Cloud@Customer and Oracle Dedicated Region Cloud@Customer are outstanding. The key benefits are migration to full cloud consumption models and the outsourcing of technical and operational skills to run the environment.

Also, many of the administrative tasks such as provisioning, configuring, tuning, encryption, security patching while running, elastic scaling, backup, failover, and recovery are automated. Organizations can also outsource some parts of SLA responsibilities and will outsource keeping the IT environment up-to-date and keeping the data safe. Automated patching removes the need for manual labor and because of this eliminates manual errors.

Another expected area of improvement is the speeding up of run times with Autonomous Databases. As the most expensive part of the total system is the database, improving elapsed times of applications reduces the number of licenses needed. Also, the database licenses include access to almost all of the database features, which can further minimize database license costs. Additional benefits of the broader range of features include improving developer productivity and expanding application functionality.

In the longer term, the economies of scale will allow Oracle to enhance availability, recoverability and security. Also, the time-to-market for application enhancements should improve as more robust change-management processes evolve. These business improvements will be significantly higher in value than the IT operational savings.

Wikibon recommends that organizations shift resources away from providing operational support. Their focus should change to managing budgeting, building and deploying applications, monitoring the Oracle services, monitoring compliance and maintaining security. Most important, Wikibon believes that Cloud@Customer allows organization management to focus on more strategic business priorities than patching, provisioning,and tuning databases.

Overall conclusions

Wikibon concludes that for organizations with large-scale mission-critical applications requiring a Tier-1 database, both Oracle Autonomous Database on Exadata Cloud@Customer and Dedicated Region Cloud@Customer are superior to cloud alternatives from AWS, Microsoft, and Google. The Oracle implementation is lower risk, has higher functionality and offers broader opportunities to redeploy or reduce internal skilled IT headcount. The latest Oracle Cloud@Customer offerings deliver an important first step in the cloud implementation of large-scale mission-critical applications. Above all else, organizations avoid the cost and risk of conversion from Oracle to alternative platforms.

Wikibon believes that Oracle is serious in migrating existing global 2000 customers to Oracle Cloud@Customer services. Success will help Oracle to develop further the automation and monitoring services, and establish better economies of scale. Wikibon believes that Oracle has a good chance to establish Cloud@Customer as the de facto standard for large-scale mission-critical Oracle Database cloud on-premises environments.

One concern that customers have expressed to Wikibon is the presenting of different cloud strategies by different Oracle sales teams. Wikibon recommends that Oracle should develop one joint overall strategic plan for migration to Oracle Cloud services with its global 2000 customers.

Action items

We believe that with this announcement, Oracle is redefining the cloud experience and breaking new ground. In our view, large Oracle customers running mission-critical applications on Oracle Databases should investigate this approach as a fundamental component of their cloud migration strategy. However, we believe it’s also critical to have a concrete plan to redeploy existing labor into more strategic projects than database task management.

Global 2000 customers running mission-critical large-scale applications on Oracle Databases should assume that they will migrate to either Autonomous Database on Exadata Cloud@Customer or Dedicated Region Cloud@Customer. Wikibon recommends that CxOs begin engaging with Oracle to conduct their TCO/ROI/IRR analyses now.

The full post at Wikibon has further Wikibon analysis and Oracle references.

Photo: SiliconANGLE