Oracle’s annual OpenWorld conference played host yesterday to the introduction of new hardware and software systems for big-data analytics. The Exalytics Business Intelligence Machine includes multiple databases to allow analytics to be run against both structured and unstructured data in memory. Oracle also announced the availability of a new SPARC-based SuperCluster for Solaris users.
The Exalytics Business Intelligence Machine is built to offer instantaneous query results on relational, multidimensional and unstructured data. Oracle CEO Larry Ellison said that it includes the TimesTen relational database for relational data, and the Essbase database for multidimensional data, both of which came to Oracle through acquisition.
Ellison also highlighted the heuristic adaptive in-memory cache. “As different people ask different questions, we migrate different data sets into memory,” he said. “If people are asking the same questions over and over again, we keep that data in memory.” He added that Exalytics can perform a logical scan on 5TB of compressed data in five seconds.
Exalytics could be seen as both a play for the burgeoning big-data market, which is currently dominated by Apache Hadoop, and also as a nod to the NoSQL movement through its inclusion of relational, multidimensional and unstructured databases in the system.
While Oracle’s two new products were the big news for software developers, Ellison spent a great deal of time discussing the advantages of using systems in which hardware and software come from the same vendor. “I remember when we first bought Sun, a lot of people said, ‘They’re going to get out of the hardware business.’ I guess we didn’t get the memo,” he said.
“A few years ago, we introduced the Exadata Database Machine, followed by the Exalogic Middleware Machine. If you design the hardware and software in concert, you can do a better job than if one company comes up with the operating system, and another comes up with the virtual machine and another creates the database.”
Thus, Ellison said, Oracle’s dedicated systems have evolved to be parallel in multiple ways. From parallel connections between the storage and database systems to parallel InfiniBand connections to the middleware machines, he claimed that parallelism on a grander scale yields simpler and more powerful systems to maintain.
“Apple is doing a good job designing software and hardware and services that work together,” said Ellison. “We said, ‘Hey that works for consumers and delivers a lot of value!’ The idea of engineering the hardware and the software—all of it—so it works together should give us the opportunity to deliver better performance and better ease of use across the board if we engineer all the separate pieces to work together. That’s the whole idea behind Exadata and Exalogic and all these engineered systems.”