IBM Looks to Bring Analytics to Where the Data Is

In a move to create to a unified data system designed to offer advanced data science capabilities across private, public or hybrid clouds, IBM at Strata Data this week announced the IBM Integrated Analytics System.

The analytics system comes with a variety of built-in data science tools designed to allow data scientists to develop and deploy their advanced analytics models in-place, directly where the data resides for greater performance.  It also supports a wide range of data types and data services, including Watson Data Platform, IBM Db2 Warehouse On Cloud, Hadoop and IBM BigSQL.

Taking analytics to the cloud

“Companies still struggle to move things to the public cloud,” said Rob Thomas, general manager for IBM Analytics, who points out that “80 percent of data is still on mainframes.” However, because the Integrated Analytics System is based on the IBM common SQL engine, it’s now easier to move workloads to the public cloud where they begin automating their businesses with machine learning. You can move and query data across multiple data stores.

[ Related: With Apache Spark, Old Mainframes Learn New Tricks ]

The Integrated Analytics System is based on IBM Data Science Experience, a set of development tools for working with Python, R, Scala and machine learning capabilities for building intelligent applications, according to Thomas. By including  Apache Spark, the analytics system brings in-memory data processing and analytics capabilities to where the data is, he said.

“Data science is a team sport,” said Thomas, and the Data Science Experience provides a set of data science tools and a collaborative work space where data scientists can create new analytic models that developers can use to build intelligent applications.

Testing, deployment and training in-place

IBM says that new to Integrated Analytics System are the machine learning capabilities that come with both its Data Science Experience and Spark embedded on the system. Because machine learning processing is embedded data does not need to be moved to the analytics processing. Testing, deployment and training can all be done in-place.

[ Related: Why Apache Spark Is So Hot ]

IBM’s Integrated Analytics System is designed to provide built-in data virtualization and compatibility with Netezza, Db2 and IBM PureData System for Analytics.

Among these capabilities, the new system also incorporates hybrid transactional analytical processing (HTAP). HTAP runs predictive analytics, transactional and historical data on the same database at accelerated response times.