Diseño e implementación de un sistema de gestión de aprendizaje institucional para la Nube. Una plataforma capaz de adaptarse dinámicamente según la demanda
Loading...
Date
Authors
Céspedes Ortiz, Job Natán
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Los sistemas de gestión de aprendizaje convencionales (LMS por sus siglas en
inglés) han sido desarrollados siguiendo los modelos de arquitectura tradicional en
infraestructura tecnológica, donde la asignación de recursos se limita y se fija según las
prestaciones del hardware utilizado. Por otro lado, han surgido alternativas disruptivas
que utilizan el paradigma de la computación en la Nube para sobrepasar ese tipo de
limitaciones. El modelo de la Nube ofrece un entorno de características que facilitan el
desarrollo de soluciones elásticas. Bajo esas condiciones una plataforma tradicional
estática se puede convertir en una dinámica con la capacidad de responder a variaciones
en la demanda y auto servirse sólo de los recursos necesarios.
La plataforma de aulas virtuales institucional de la Universidad de Costa Rica
utiliza Moodle como LMS; la propuesta actual para su arquitectura permite robustez para
soportar el servicio, pero sigue el modelo de asignación de recursos tradicional. La
Universidad tiene también un proyecto de Nube Académica que posibilita el desarrollo de
soluciones elásticas en un entorno basado en el gestor de Nube OpenStack. El objetivo ha
sido integrar este concepto de elasticidad en el diseño e implementación de una nueva
plataforma que utilice Moodle como LMS, aprovechando las características de la Nube
para obtener un grado de adaptabilidad dinámica.
Para lograrlo se identificaron, detallaron y diseñaron los componentes base de una
arquitectura auto escalable para un LMS, según las características del problema, los
requerimientos y los recursos disponibles. El diseño propuesto sirvió como insumo para
implementar un prototipo en el entorno de la Nube disponible. Posteriormente se
diseñaron, aplicaron y analizaron dos experimentos para evaluar la capacidad de
adaptación dinámica del prototipo respecto a la demanda y para valorar su desempeño.
El resultado fue un prototipo elástico para el LMS Moodle, que es además tolerante
a fallas y altamente disponible. Fue evidente a partir de la evaluación que el prototipo
tiene un rango de adaptabilidad. Esta característica le permite ajustar su capacidad de
forma dinámica ante cambios en la demanda, utilizando diferentes estrategias para agregar
o liberar recursos. Aunque la variabilidad en los tiempos de respuesta y el consumo de
recursos en el hipervisor durante la evaluación se incrementaron con la característica
elástica; el prototipo tuvo un buen rendimiento respecto a las otras arquitecturas
comparadas. La solución sirve como base, no solo para Moodle, sino también para otros
sistemas tradicionales similares, que tienen en la Nube una forma para optimizar el uso de
los recursos computacionales, disminuir los tiempos de implementación de las soluciones
y aumentar las facilidades en las tareas de gestión de sus servicios.
Conventional learning management systems (LMS) have been developed following the traditional architecture models in technological infrastructure, where the allocation of resources is limited and set according to the features of the hardware used. On the other hand, disruptive alternatives that use the cloud computing paradigm have emerged to overcome these kind of limitations. The Cloud model offers an environment with features that make the development of elastic solutions easier. Under these conditions, a traditional static platform can become a dynamic one with the ability to respond to demand variations and serve itself only with the necessary resources. The University of Costa Rica's institutional virtual classrooms platform uses Moodle as LMS; the current model for its architecture allows robustness to withstand the service, but it follows the traditional allocation of resources. The University also has an Academic Cloud project that enables an environment based on OpenStack Cloud manager to develop elastic solutions. The aim has been to integrate the elasticity concept in the design and implementation of a new platform, in a way that it continues using Moodle as LMS, takes advantage of the Cloud features and obtains a degree of dynamic adaptability. To achieve this aim, the basic components for an auto scalable architecture were identified, detailed and designed, according to the characteristics of the problem and the existing requirements and resources. The proposed design provided an input to implement a prototype in the Cloud environment available. After that, two experiments were designed, performed and analyzed to evaluate the capacity of the prototype for dynamic adaptation in relation to the demand and also to evaluate its performance. The result was an elastic prototype for a LMS platform based on Moodle that is also fault-tolerant and highly-available. From the evaluation, it was evident that the prototype has an adaptability range. This characteristic allows to adjust its capacity in a dynamic way, following different strategies to add or release resources when faced with changes in the demand. Even though the variability in response times and in the hypervisor consumption of resources increased with the elastic feature; the prototype performed well in relation to other architectures compared. The solution serve as a basis not only for Moodle but also for other similar traditional systems that have in the Cloud a way to optimize the use of computing resources reduce implementation times and increase facilities in managing tasks of their services.
Conventional learning management systems (LMS) have been developed following the traditional architecture models in technological infrastructure, where the allocation of resources is limited and set according to the features of the hardware used. On the other hand, disruptive alternatives that use the cloud computing paradigm have emerged to overcome these kind of limitations. The Cloud model offers an environment with features that make the development of elastic solutions easier. Under these conditions, a traditional static platform can become a dynamic one with the ability to respond to demand variations and serve itself only with the necessary resources. The University of Costa Rica's institutional virtual classrooms platform uses Moodle as LMS; the current model for its architecture allows robustness to withstand the service, but it follows the traditional allocation of resources. The University also has an Academic Cloud project that enables an environment based on OpenStack Cloud manager to develop elastic solutions. The aim has been to integrate the elasticity concept in the design and implementation of a new platform, in a way that it continues using Moodle as LMS, takes advantage of the Cloud features and obtains a degree of dynamic adaptability. To achieve this aim, the basic components for an auto scalable architecture were identified, detailed and designed, according to the characteristics of the problem and the existing requirements and resources. The proposed design provided an input to implement a prototype in the Cloud environment available. After that, two experiments were designed, performed and analyzed to evaluate the capacity of the prototype for dynamic adaptation in relation to the demand and also to evaluate its performance. The result was an elastic prototype for a LMS platform based on Moodle that is also fault-tolerant and highly-available. From the evaluation, it was evident that the prototype has an adaptability range. This characteristic allows to adjust its capacity in a dynamic way, following different strategies to add or release resources when faced with changes in the demand. Even though the variability in response times and in the hypervisor consumption of resources increased with the elastic feature; the prototype performed well in relation to other architectures compared. The solution serve as a basis not only for Moodle but also for other similar traditional systems that have in the Cloud a way to optimize the use of computing resources reduce implementation times and increase facilities in managing tasks of their services.
Description
Keywords
Computación en la Nube, LMS, Autoescalabilidad, 004.678 Internet
Citation
http://pci.ucr.ac.cr/trabajos-de-graduaci%C3%B3n/dise%C3%B1o-e-implementaci%C3%B3n-de-un-sistema-de-gesti%C3%B3n-de-aprendizaje-institucional