Oracle vice president of product management Neil Mendelson told iTWire that the Oracle Big Data Appliance was introduced over two years ago "priced at the commodity side of the business" to give customers an alternative to building their own Hadoop clusters.
But Big Data Appliance owners have moved from experimentation to production systems, and that means they need to use the appliance within the framework of their other data management systems.
In most situations, organisations are using Hadoop alongside traditional data warehouses, explained principal product manager Dan McClary, but they want to avoid "the big data skills gap," creating another data silo, or running into security problems.
SQL is widely known; Oracle Big Data SQL allows one SQL query to run across Hadoop, NoSQL and Oracle Database; and it also allows existing security policies and frameworks to be extended to Hadoop.
From a database administration, security management or reporting standpoint, "this all works just like Oracle Database," he said.
Furthermore, Big Data SQL means SQL-based business intelligence and other tools can easily access data from Hadoop and NoSQL sources.
"It's really about what you don't have to do," said Mr Mendelson. "The best solution is the simplest solution."
Page 2: "all the magic happens in the background."
The broad adoption of big data has been limited by the absence of declarative access, he said, but Oracle is providing users with the power to use SQL on data wherever it lives.
"We think this is going to be appealing to the industry at large," said Mr McClary.
To maximise performance, data remains on the source system as far as possible - for example, only the result of an SELECT statement is transferred. Applying the Smart Scan concept from Exadata to Hadoop provided "a massive improvement," he said. Data from Hadoop and NoSQL is presented as external tables.
Mr McClary gave the example of Twitter data, which can be sensibly stored in Hadoop. While it can be analysed there, the question for many organisations is "how do we bring that data to the data warehouse?" - perhaps to relate Twitter sentiment and purchases.
More generally, it provides a way of efficiently storing new types of data (eg, storing sensor logs in Hadoop) while making sure it is accessible to enterprise users alongside traditional data sets.
You no longer need to worry about "uncurated data" - it can be stored wherever seems appropriate and without cluttering up the data warehouse, yet it can be drawn upon via SQL as necessary.
Big Data SQL will be available in the current quarter, following the release of Oracle Database 12.1.02. It will initially require Oracle Database 12c on Exadata, but will spread to other Oracle Engineered Systems and then to other systems running Oracle software.