Publication Date: 2023/12/14
Abstract: In massive data processing, platforms using MapReduce are designed for data centers, which are generally centralized.These platforms typically rely on a single node to maintain and coordinate MapReduce tasks, leading to a single point of failure. Our aim in this paper has been to propose a model for MapReduce computation on the Red Green Blue architecture, which is a decentralized, triple-node big data architecture. This architecture is based on the peer-to-peer networking protocol named Content Addressable Network. First, we implemented all the steps of the MapReduce computation approach, taking into account the properties of the Content Addressable Network protocol and the Red Green Bluearchitecture. We then carried out an experiment in a local network to evaluate performance in terms of processing speed and time. The experiment showed that latency decreased with the number of compute nodes. This study not only showed that the Red Green Blue architecture is viable as a massive data processing architecture, but also improved processing times as a function of network nodes. The robustness, scalability and lack of a single point of failure of the Red Green Bluearchitecture mean that MapReduce can be easily deployed in a wider variety of applications.
Keywords: P2P protocol, Map Reduce, RGB architecture, Big data Storage.
DOI: https://doi.org/10.5281/zenodo.10375062
PDF: https://ijirst.demo4.arinfotech.co/assets/upload/files/IJISRT23DEC313.pdf
REFERENCES