Intel has CPU encryption enabled for Hadoop

Chip giant Intel is redoubling its efforts to defend its valuable data-center territory-specifically to develop its own technology to drive data management and analytics technologies, such as hadoop--implementations.

To ensure that the Xeon chips are the preferred platform for running a large Hadoop cluster in the data Center Administrator's account, Intel announced in Tuesday that it will be the Intel distribution for Apache Hadoop (its own, Open-source software derivative solution) Add a number of new features and technologies.

This round of updates contains the second version of the Intel Graph Builder for Apache Hadoop, the Intel Analytics Toolkit for Apache Hadoop, and the Intel Expressway Tokenization Boker.

The "Rhino Project" is particularly noteworthy among the many major Intel projects dedicated to Hadoop, and it is dedicated to providing a framework that leverages the x86 AES processor directive to provide hardware-accelerated encryption and decryption for Hadoop. The project was designed to respond to the recent spate of Snowden events, in the hope of overcoming the reliability crisis in the x86 rdrand operation of a well-known chipset that has been successfully cracked by the US National Security Service (FreeBSD). OpenSSL also specifically addressed this issue.

But to build this Hadoop release, Intel has "enabled additional encryption in HBase," Ritu Kama, head of product management at Intel's Big Data business unit, said in an interview. These features "Enable transparent encryption of hbase tables and columns while extending the encryption mechanism in HBase to the cell level." ”

This package is 20 times times faster than using software on the same hardware stack, Kama told us.

Other new features include the Intel Analytics Toolkit, which is designed to help staff with data access a set of algorithms and machine learning patterns.

"We are developing a complete set of artifacts or algorithms that will allow users to create applications directly from the toolkit-whether or not they are clustered in a way that is recommended," Kama.

"You don't have to start from scratch every time. We will provide a set of procedures to guide users to place data under the input directory. The format of the data can be varied--log files, structured or unstructured ... Then we'll help the user organize the data into a standardized format in accordance with the process, so that the algorithm can be used, "she explains.

Looking to the future, Intel "may also provide a programming environment or IDE integration solution that developers can use to visually drag and drop to implement data import," she says.

In addition to this toolkit, Intel has published "graph Builder", designed to help administrators successfully accept the data stored by Hadoop and aggregate the results into graphical form-"Retailers can create graphical results based on information compiled from their historical sales data and social media data, To better understand the real relationship between brand appeal and customer buying habits, "explained the Intel side in a recording statement."

Intel is devoting a lot of effort to Hadoop-related projects because the chip giant feels the platform is about to become one of the core software systems for data processing. In addition, Intel wants to ensure that its own chip products maintain a leading edge with rivals such as AMD. For this reason, Intel has adopted a large number of open source technologies in its own Hadoop project, with the exception of a holistic "Intel Hadoop Manager" layer.

"We are not going to really build a huge, exclusive intellectual property system," explains Jason Fedder, director of marketing and business operations at Intel's data center software department. "Our focus is on creating a debug-optimized component solution that accelerates the actual performance of our core Xeon product lines in the data center." ”

The analysis kit will be launched in the first quarter of 2014, and the base price will be announced by Intel by then. The Graph Builder Toolkit will be published next January as an open-source downloadable form.

As for Intel's distributors, which include the management side, the price per node is between 1500 and 3300 dollars, "depending on the total number of nodes and the actual support scheme (seven days a week, 24 hours a day, or Five days per week, nine hours a day)," A spokesman for Intel Corporation told us in an email.