Advances in Computational Intelligence and Communication by Unknown

Author:Unknown
Language: eng
Format: epub, pdf
ISBN: 9783031195235
Publisher: Springer International Publishing

2.2 Research on the Evolution of Architectures

Evolutions have been made to analyze business data. In the context of process monitoring, M.F. Huber processes a large amount of complex events using Spark job programming [5]. Spark CEP is a stream processing engine on top of Apache Spark supporting continuous query language. In this use case, the implementation of a data lake is based on Apache Cassandra, a column-oriented database. This work shows that it is possible to change the architecture, but this requires recreating a new monolithic solution from the persistence layer. The choice of Cassandra instead of HBase is explained by the structured nature of the stored data. Typing is more important when analyzing and building predictive models. M. Y. Santos uses the Hadoop stack in his study and insists on the constraints imposed in particular on the Spark streaming versions and more specifically the Java 8 language [6]. The information coming from the devices are routed with REST services. The authors have chosen Cassandra as a NoSQL database whose version is linked to that of Spark streaming. An update of Java to an even slightly higher version is not possible.

It is obvious that any assembly of strongly coupled software is not scalable. It is necessary to turn to new software architectures. T Ivanov proposes an abstract architecture in the manner of the Model-Driven Architecture (MDA) description [7]. However, this is only an anonymization of a current architecture with still strong dependencies between components. M. S. Hadj Sassi has published about a work on an architecture for the collection of data from sensors and their analysis [8]. He focuses his solution on the definition of business services specific to the distribution of sensors, and he adopts above all an asynchronous coupling between services. Throughout the paper, he presents a business workflow based on this principle where each phase is decomposed into a component diagram illustrating the asynchronous points.

G. Moktari followed the same approach for the definition of a layered architecture in the context of Big Data applications for the smart home [9]. The author insists mainly on the real-time aspects even if some asynchronous points are put forward, for example, at the end of the data collection. As the paper describes a case study where the information comes from buildings with sensors, the data collections are done via a cloud where the Big Data processing takes place. The use of REST APIs indicates that the real-time aspects are not preserved, but the layered architecture underlines a geographical division of the processing and necessarily the isolation of some software. Through these works, we can see software properties that have become important because Big Data technologies are evolving rapidly. In addition, the basics of use have also become more complex with an increasing amount of data and access requirements that need to be continuously improved.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.