It's no longer news that Oracle has gotten cloud religion. While not the first to market, Oracle has now been in the cloud market long enough to get within eyeshot of having the full buildout of its global presence and portfolio of services. As one of the established names in the data center, the cloud has turned the tables and placed Oracle in the position of challenger, and not surprisingly, Amazon is dead center in its sights.
Oracle's positioning is that its later start to the cloud has provided the chance to learn all the lessons of the first-generation providers. Although not using those exact words, Oracle is framing its platform as Cloud 2.0.
Having the benefit of a later start, Oracle has had the luxury of planning out, rather than retrofitting the building blocks of its platform. From the bottom up, Oracle's cloud platform is a series of pieces that are added together; from the top down, it looks like a Russian doll. Oracle's Infrastructure as a Service (IaaS) is incorporated into its Platform as a Service (PaaS) offerings. And in turn, the Software as a Service (SaaS) incorporates the infrastructure and APIs of IaaS along with the database and middleware of the PaaS platform. Examples of the benefits of this approach include single sign-on across the whole SaaS and PaaS portfolio of services and leveraging of the integration PaaS services to support drag-and drop-application integration with the SaaS cloud.
Oracle rethought implementation of cloud security by situating most of the defenses, not at the perimeter of the data center, but out at the edge -- which would be the point of presence or at peering exchanges like Equinix. The guiding notion is to keep malware and bots as far away from the cloud data center as possible.
It has also rethought networking. While it is not alone in having high-speed backbones linking its cloud data centers, within the data centers, its new design is built around higher-speed 25 gigabit Ethernet links. The network topology of its cloud data centers is extremely flat with no more than two hops separating any node. With the high speed, flat topology, Oracle is positioning its public cloud to deliver higher performance and firm SLA guarantees.
It also offers a simplified connect to its IaaS services with a single API, meaning the interfaces are the same, regardless of whether you are running a non-virtualized bare metal compute service or conventional multi-tenant deployment.
Oracle's autonomous database 18c, which should be in production by the end of this month, sets the stage for how Oracle intends to use AI to not simply automate cloud services, but eliminate human intervention in tuning or configuring them.
The first shot was fired with announcement of database 18c back at Oracle OpenWorld last fall. With its cloud-first strategy, Oracle Autonomous Data Warehouse is being released initially on the Oracle Cloud, and on-premises with Exadata Cloud at Customer. Functions such as storage management, query optimizations, query rewrites, and database diagnostics are performed automatically. Autonomy also means eliminating most, if not all, of the decisions on how to tune or configure the database. Beneath the hood, the autonomous database uses machine learning rather than rules to determine how to partition, index, and secure data; conduct self-healing; and perform backups and disaster recovery. Later this year, Oracle will release autonomous database cloud services for transaction systems, along with Oracle NoSQL database and Graph analytics. We expect within the year to see rival cloud platforms respond with adding machine learning (superseding rules) to automate their databases.
While the upcoming autonomous database releases round out the data platform product line, we believe that Oracle will ultimately extend autonomous operation to the rest of its PaaS portfolio from analytics to content management, integration, IoT, API management and other services. Oracle is also beginning to infuse AI into its enterprise applications. There are clear opportunities with applications such as customer engagement, where functions such as identifying a customer and generating next-best offer are horizontal processes that are quite common across companies. On the other hand, defining fraudulent transactions can vary by industry, meaning that the onramp for AI in organization-specific areas will likely be longer and require more sophisticated machine learning or deep learning algorithms.
The Oracle Cloud competes in a very asymmetrical market: the lineup of players differs depending on whether you are talking about IaaS, PaaS, or SaaS. Oracle may compete in IaaS with a newer generation architecture, but Amazon, followed by Microsoft, have mindshare and (for now) a broader array of services.
For PaaS, Oracle's prime rivals are Microsoft and IBM, and in selected PaaS segments, Amazon and the Salesforce platform. Given the diversity of PaaS, the competitive landscape differs by service. For instance, in machine learning, Amazon, Microsoft, Google, and IBM already have services, while Oracle's won't debut until late winter or spring. Conversely, in emerging areas like blockchain, virtually all cloud players are still at the starting gate.
In SaaS, the landscape amounts to a long tail, comprised of a diverse array of established ERP vendors such as SAP, and specialized cloud-only providers (headlined by Salesforce, which pioneered SaaS and boasts the largest third-party ecosystem). Oracle counters with an offering that has far more breadth in enterprise applications. There is clearly a synergy between having a broad SaaS portfolio for enterprise heartbeat applications and cloud infrastructure designed for high service levels.
So, it's not surprising that SaaS is the hot spot of Oracle cloud portfolio, with revenues having grown 55% in the most recent quarter, compared to 44% for the cloud overall, and 6% for the entire company. Putting the cloud numbers in perspective, as of the most recent quarter, Oracle's $1.5 billion cloud business is roughly a third of the size of Amazon's, which came in at $4.58 billion in Q3.
Oracle has not restricted its public cloud to third parties. It has a marketplace; features a number of services based on open source technologies such as Kubernetes and Kafka; and is drawing a growing roster of SI partners. Nonetheless, the brunt of Oracle's energy has been in optimizing its cloud as the most economic and efficient place to run Oracle software.
For Oracle 18c data warehousing customers, Oracle promises to underprice Amazon Redshift and Oracle on Amazon RDS by half. This came months after Oracle doubled pricing for running its database on Amazon's RDS service. Oracle claims its database is cheaper to run than Amazon, period, a point that chairman Larry Ellison was hardly bashful about driving home during his OpenWorld keynote. Not surprisingly, Amazon wasn't afraid to dish it back.
What makes Oracle's promise audacious is that it extends beyond its own database on RDS to Redshift, where the price is set by Amazon. Given that Amazon's databases have lately been its fastest growing products -- Aurora now, and Redshift before that -- that sets up a virtuous cycle of competition in price and innovation for customers who are pondering their cloud database choices.