Apache Hama

Apache Hama is a toolkit for usage in mediums where large quantities of data need to be analyzed and processed.

Hama specifically targets scientific mediums, being able to also interpret and execute basic or more advance mathematical operations, while also functioning from multiple clusters (cloud-based infrastructures).

This is mainly thanks to its BSP (Bulk Synchronous Parallel) features, being able to handle anything from matrix-based data, to network-based computations and many other more.

As for its architecture Hama is comprised of three parts: the shell, the Hama core, and a public API.

These tools can be used to interact with the system either in real-time or for scheduled tasks.

What is new in this release:

• Fix/improve DoubleVector and DenseDoubleVector
• Added persistent queue option to JobConf
• Support NullWritable in Hama Pipes
• Added sort behaviour to runtime partitioner
• Added NamedVector
• Added feature transformer interface to improve the power and flexibility of existing machine learning model
• CompositeInputFormat in Hama

What is new in version 0.6.4:

• Fix/improve DoubleVector and DenseDoubleVector
• Added persistent queue option to JobConf
• Support NullWritable in Hama Pipes
• Added sort behaviour to runtime partitioner
• Added NamedVector
• Added feature transformer interface to improve the power and flexibility of existing machine learning model
• CompositeInputFormat in Hama

What is new in version 0.4.0:

• Stabilized barrier synchronization.
• Multiple tasks per node.
• Input/Output formatter.
• Message Combiners.
• Improved examples.
• Benchmark test results.

What is new in version 0.2.0:

• BSP computing framework and its examples.
• Pi Estimator and Serialize Printing.
• CLI-based managing and monitoring tool of BSP job.

Requirements:

• Java 1.6 or higher
• Apache Hadoop 1.0 or higher