Nt1330 Unit 4 Case Study And Data Analysis Essay

Words: 553
Pages: 3

a. Google’s MapReduce:
The original MapReduce implementation by Google is developed for large clusters of networked machines. The MapReduce library handles parallelization and data distribution. MapReduce is designed mostly for the developers needn’t worry about parallel and network programming problems and they focus on the real problems.
b. Hadoop:
Hadoop is a MapReduce implementation by Apache. Data is distributed across the machines in a network using the Hadoop Distributed File System (HDFS). Hadoop can be organized to run multiple simultaneous Map tasks on single nodes. In multi-core systems, this is a great benefit, as it allows making full use of all cores.
c. GridGain:
GridGain is an open-source MapReduce implementation. In the MapReduce algorithm, the task is split into subtasks and workers drag the split parts as soon as they have free processor time. In GridGain, the subtasks are pushed to the nodes. The extra functionality it introduces some additional complexity the developer has to plan in advance so that no worker does stay without reason idle.
…show more content…
Phoenix:
Phoenix is a MapReduce implementation intended for shared-memory systems. Phoenix can be used to parallelize data intensive computations on multicore and multiprocessor computers. The principles in Phoenix implementations are essentially the same as in original MapReduce, except instead of large clusters, it is intended for shared-memory systems. Overheads caused by task spawning and data communications can be minimized when working in a shared-memory environment.
e. Mars:
Mars implements the MapReduce framework for graphics processors (GPU). The purpose of the Mars framework is to hide all the complex details of the GPU.
f. Map-Reduce-Merge:
Map-Reduce-Merge is an extension to the MapReduce programming model, rather than an implementation of MapReduce. MapReduce programming model does not directly hold processing multiple related heterogeneous