Oracle Automates the Data Warehouse
Oracle Corp. released the first of what it says will be a string of cloud platform services based on machine learning that are designed to differentiate the cloud upstart from public cloud giant Amazon Web Services (NASDAQ: AMZN).
Larry Ellison, Oracle’s CTO and board chairman, wasted no time during a company event this week laying out “our strategy versus Amazon,” announcing general availability of its “autonomous data warehouse cloud.” The overarching goal is reducing costs, speeding application development and boosting database security via automated services. Ellison said the approach moves beyond the AWS infrastructure-as-a-service model to deliver “a complete suite of platform services that are at a higher level than low-level infrastructure stuff.”
“Our strategy going forward is to develop a high-level layer of software called PasS services that sit on top of infrastructure,” Ellison said Tuesday (March 27). “Rather than developing database applications like you used to develop, you’ll have a new set of tools for application development [of] database services that are not business as usual….” Instead, “you’ll be using these very high-level services that allow you to build applications much faster.”
Oracle (NYSE: ORCL) announced is autonomous database approach last fall. The platform in beta testing since then is now generally available. Seldom one to understate his case, Ellison said Oracle’s autonomous database “is probably the most important thing we have ever done” in data management.
“We need automated systems, intelligent systems to protect our data,” he added.
Oracle makes that case that autonomous databases and cyber security are emerging machine learning applications. “We are using machine learning to make our software smarter, to make our software ‘self-driving’,” Ellison said. Among the goals is harnessing the ability of machine learning models to spot patterns in huge data sets and make reliable predictions as models are trained.
In the case of the autonomous database, its crunches log-in data, for example, to spot anomalous data. In another example, machine learning could help spot anomalous SQL queries to a database. “Now this is all automated,” Ellison noted, including security capabilities like intrusion detection as well as databased provisioning, tuning and upgrades.
In its drive to differentiate itself from AWS and other cloud vendors, Oracle is promoting services such as machine learning tools that can, for example, automatically track event logs, detect a database intrusion and automatically patch itself to prevent system downtime.
Those features distinguish Oracle’s automated database from the AWS “elastic” cloud and its Aurora and Redshift databases, Ellison claimed. “Can they automatically expand as you need more capacity? No, not ‘elastic’ at all,” he continued.
Ellison claimed last fall that Oracle Cloud was more than 100 times faster for database analytics than Amazon Redshift. The broadside was part of a larger Oracle campaign aimed at challenging public cloud leader AWS with an expanded analytics package that includes applications, infrastructure and database analytics
Now, Oracle is stressing platform services and automation as a way of cutting costs associated with the labor-intensive management of databases. Automation also helps reduce costs by eliminating human error while boosting reliability and availability. “If you eliminate human labor, you eliminate human error,” the Oracle co-founder asserted last fall when he launched his cloud competition with AWS.
For its part, Amazon (NASDAQ: AMZN) has dismissed Ellison’s broadsides as bluster while market analysts take a wait-and-see approach on whether Oracle can deliver. Meanwhile, Amazon and Salesforce (NYSE: CRM) are reportedly moving off Oracle databases in response to Ellison’s broadsides. Oracle disputes this.
Nevertheless, AWS has unveiled database options built on open source software. Among them is a new graph database dubbed Nepture, that complements a growing list of AWS platforms, including relational, NoSQL and in-memory databases along with key-value and object stores.