Intel has launched Data Platform, a suite of cloud-based products and services designed for developing BI systems.
The open platform is based on the Intel Distribution of Hadoop, with the addition of streaming data processing, graph processing, and interactive and iterative data analysis. There’s also an analytics toolkit that provides the graph analytics and predictive modeling.
The Intel Distribution includes the Apache Hadoop framework, MapReduce, HDFS, and other related components to support both batch processing and near-time analytics, including the Apache Hive data warehouse infrastructure, Apache Pig data flow language, and Apache HBase database.
The platform is currently in beta, and will be available next quarter in two versions – Enterprise Edition and Premium Edition. Enterprise Edition will be free but won’t include product support. Premium Edition will be available subscription based and will provide additional technical features including enhanced automation, proactive security fixes and alerts, ongoing feature enhancements, and live telephone technical support.
The toolkit consists of a set of data analysis algorithms, such as graphs and network-based clustering, that IT teams can build on and that can be customized with code. The toolkit also includes a development framework for unifying graph analytics and classical machine learning, according to the announcement about the new suite.
Intel's announcement suggests that “using the toolkit, a data scientist at a financial services firm can gain richer insight by developing a fraud detection service that identifies patterns between purchasers, merchants and transactions to uncover potential points of compromise instead of only monitoring the purchase habits of individuals.”
Boyd Davis, Intel vice president and general manager of the its Datacenter Software Division, said
“Much like an operating system for big data processing, the Intel Data Platform supports a wide variety of applications."
While the announcement says the platform is in beta, there are as yet no obvious links to it on the Intel website, nor any references to it on the Intel blog.