By Jean-Pierre Dijcks-Oracle on Apr 14, 2015
Click here to start reading the Full Statement of Direction.
Introduction: Oracle Big Data Management System Today
As today's enterprises embrace big data, their information architectures must evolve. Every enterprise has data warehouses today, but the best-practices information architecture embraces emerging technologies such as Hadoop and NoSQL. Today’s information architecture recognizes that data not only is stored in increasingly disparate data platforms, but also in increasingly disparate locations: on-premises and potentially multiple cloud platforms. The ideal of a single monolithic ‘enterprise data warehouse’ has faded as a new more flexible architecture has emerged. Oracle calls this new architecture the Oracle Big Data Management System, and today it consists of three key components
- The data warehouse, running on Oracle Database and Oracle Exadata Database Machine, is the primary analytic database for storing much of a company’s core transactional data: financial records, customer data, point- of-sale data and so forth. Despite now being part of a broader architecture, the requirements on the RDBMS for performance, scalability, concurrency and workload management are in more demand than ever; Oracle Database 12c introduced Oracle Database In-Memory (with columnar tables, SIMD processing, and advanced compression schemes) as latest in a long succession of warehouse-focused innovations. The market-leading Oracle Database is the ideal starting point for customers to extend their architecture to the Big Data Management System.
- The ‘data reservoir’, hosted on Oracle Big Data Appliance, will augment the data warehouse as a repository for the new sources of large volumes of data: machine-generated log files, social-media data, and videos and images -- as well as a repository for more granular transactional data or older transactional data which is not stored in the data warehouse. Oracle’s Big Data Management System embraces complementary technologies and platforms, including open-source technologies: Oracle Big Data Appliance includes Cloudera’s Distribution of Hadoop and Oracle NoSQL Database for data management.
- A ‘franchised query engine,’ Oracle Big Data SQL, enables scalable, integrated access in situ to the entire Big Data Management System. SQL is the accepted language for day-to-day data access and analytic queries, and thus SQL is the primary language of the Big Data Management System. Big Data SQL enables users to combine data from Oracle Database, Hadoop and NoSQL sources within a single SQL statement. Leveraging the architecture of Exadata Storage Software and the SQL engine of the Oracle Database, Big Data SQL delivers high-performance access to all data in the Big Data Management System.
Using this architecture, the Oracle Big Data Management System combines the performance of Oracle’s market-leading relational database, the power of Oracle’s SQL engine, and the cost-effective, flexible storage of Hadoop and NoSQL. The result is an integrated architecture for managing Big Data, providing all of the benefits of Oracle Database, Exadata, and Hadoop, without the drawbacks of independently-accessed data repositories.
Note that the scope of this statement of direction is the data platform for Big Data. An enterprise Big Data solution would also be comprised of big data tools and big data applications built upon this data platform.