Aster In-Database MapReduce
MapReduce has been discussed at length at InfoQ, and is a programming model originally introduced by engineers at Google as a scalable approach to processing large data-sets.
The nCluster database is labeled by Aster as a massively parallel processing (MPP) database. The parallel architecture of nCluster is described on their website in this way:
Aster nCluster is built on a unique, multi-tiered nCluster architecture which consists of three separate classes of nodes: Queens, Workers, and Loaders. The three-tier design encapsulates a clean separation of roles for analytic processing. Each tier can be independently and incrementally scaled in response to the workload characteristics – adding more capacity (Workers), loading bandwidth (Loaders), or concurrency (Queens) on an as-needed basis.The MapReduce implementation provided in Aster nCluster allows for the execution of MapReduce calculations within the database, using this same architecture:
Just like its massively parallel execution environment for standard SQL queries, Aster nCluster now adds the ability to implement flexible MapReduce functions for parallel data analysis and transformation inside the database. Aster nCluster In-Database MapReduce functions are simple to write and are seamlessly integrated within SQL statements. They rely on SQL queries to manipulate the underlying data and provide input. The functions can procedurally manipulate such input data and provide outputs that can be further consumed by SQL queries or be written into tables within the database.SQL/MR is a special SQL MapReduce function library introduced by Aster that can be used to invoke map-reduce algorithms within the nCluster platform. Aster supports polymorphic functions and dynamic typing, and MapReduce calculations may be developed in languages such as Java, Python, C++ and others.
More information about In-Database Map Reduce and the nCluster database is available on the Aster Data Systems website.
Stephanie Davis (nee Stewart) Dec 21, 2014