Design and Evaluation of Network-Levitated Merge for Hadoop Acceleration

Hadoop is a popular Interchange document implementation of the Map scale back programming model for cloud computing. However, it faces variety of problems to know the foremost effective performance from the underlying systems. These embrace a business barrier that delays the scale back section, repetitive merges and disk accesses, and together the dearth of quality to whole fully totally different interconnects. to stay up with the increasing volume of datasets, Hadoop additionally needs economical I/O capability from the underlying portable computer systems to methodology and analyze information. we've got Associate in Nursing inclination to elucidate Hadoop, degree acceleration framework that optimizes Hadoop with plug-in components for quick information movement, overcoming these limitations. a singular network-levitated merge rule is introduced to merge information whereas not repetition and access. additionally, a full pipeline is supposed to overlap the shuffle, merge and scale back phases.