Intel has introduced its own distribution of Hadoop, incorporating enterprise level features for security, performance and management. The IT giant has also announced a number of partners for its offering, which it will sell on a subscription basis, ranging from the likes of Cisco Systems and Dell, to MarkLogic, SAP and Teradata.
As part of its distribution, Intel has made software updates to a number of Hadoop components, including the HDFS file system, Yarn distributed processing framework, Hive SQL interface and HBase columnar store. These updates have been contributed back to the Apache open source project, on which Intel’s distribution is based.
Performance enhancements include optimisation for solid state disks and cache acceleration, and hardware-based encryption and decryption leveraging the AES instructions of Intel chips.
Intel has also introduced a proprietary module – Intel Manager for Apache Hadoop – which provides additional functionality for deployment, management, monitoring and security.
As well as its own Hadoop distribution, Intel is continuing to develop its Graph Builder visualisation tool for analysing Hadoop-based data. It has also made investments in 10gen and its MongoDB NoSQL database and in operational analytics specialist Guavus.
In introducing its own distribution, Intel expects to accelerate deployment of Hadoop – and sales of its microprocessors, SSDs and networking products alongside. That will mean increased competition for the big three Hadoop startups – Cloudera, Hortonworks and MapR Technologies, who all offer distributions with their own added features and functionality.