Home // International Journal On Advances in Networks and Services, volume 10, numbers 3 and 4, 2017 // View article
Authors:
Pascal Lau
Paolo Maresca
Keywords: distributed computing; high performance computing; data systems.
Abstract:
On a daily basis, Internet services experience growing amount of traffic that needs to be ingested first, and processed subsequently. Technologies to streamline data target horizontal distribution as design tenet, giving off maintainability and operational friendliness. The advent of the Internet of Things (IoT) and the progressive adoption of IPv6 require a new generation of scalable data streamline platforms, bearing in mind easy distribution, maintainability and deployment. Chimera is an ultra- fast and scalable Extract Transform and Load (ETL) platform, designed for distribution on commodity hardware, and to serve ultra-high volumes of inbound data, processing in real-time while offering a simplistic fault model. It strives at putting together top performance technologies to solve the problem of ingesting huge amount of data delivered by geographically distributed agents. It has been conceived to propose a novel paradigm of distribution, leveraging a shared nothing architecture, easy to elastically scale and to maintain. It reliably ingests and processes huge volumes of data: operating at the line rate, it is able to distribute the processing among stateless processors, which can dynamically join and leave the infrastructure at any time. Experimental tests show relevant outcomes intended as the ability to systematically saturate the I/O (network and disk), preserving reliable computations (at-least-once delivery policy).
Pages: 117 to 129
Copyright: Copyright (c) to authors, 2017. Used with permission.
Publication date: December 31, 2017
Published in: journal
ISSN: 1942-2644