Downloads: 1 | Views: 32 | Weekly Hits: ⮙1 | Monthly Hits: ⮙1
Informative Article | Data & Knowledge Engineering | India | Volume 10 Issue 6, June 2021 | Rating: 4.8 / 10
Analysis on G-Hadoop for Big Data Computing Across Distributed Cloud Data Centers
Kartheek Pamarthi
Abstract: The computational demands for scientific data processing on a wide scale have recently increased dramatically. For example, in 2010, the data created by the Large Hadron Collider (LHC) in the domain of High Energy Physics (HEP) amounted to thirteen petabytes. In 34 different locations, 140 computing centres handle this massive volume of data. In recent years, the MapReduce paradigm has been the go-to programming approach for data-intensive, enterprise-level applications. Unfortunately, large-scale distributed data processing over several clusters is not yet possible with the MapReduce implementations that are available. These implementations were created for use with single cluster scenarios. One popular MapReduce implementation that uses a cluster to execute MapReduce jobs is the Hadoop framework. With G-Hadoop, the Hadoop MapReduce framework may be extended to conduct MapReduce workloads on several clusters. G-Hadoop, in contrast, only reuses the user login and job submission system of Hadoop (which is single-cluster). A novel G-Hadoop security model is suggested in this paper. The SSL protocol and public key cryptography are two of the many security technologies upon which this architecture is based; it was designed with dispersed environments in mind. Using a single-sign-on approach, this security framework streamlines the job submission and user authentication processes of the present G-Hadoop implementation. Furthermore, the G-Hadoop system is protected from conventional threats by means of a variety of security techniques provided by the built security architecture.
Keywords: scientific data processing, MapReduce, G-Hadoop, security model, distributed environments
Edition: Volume 10 Issue 6, June 2021,
Pages: 1838 - 1845