Creation and evaluation of an educational framework for use in network teaching

Teaching and assessing students on the practical side of networking can be achieved through the use of simulators. However network simulators are limited in what can they can do, since the device being simulated is not fully functional and the generation of the exercises typically results in the same specification being presented to the student [1, 2]. When the student has finished an exercise they are typically just presented with an outline grade with little indication of areas of weakness or strength.
This thesis investigates how the Bloom [3] and SOLO [4] learning taxonomies can be used to specify and grade network challenges, while using the idea of fading worked examples [5] to design the challenges to lower the cognitive load on the student. It also proposes a framework that can be used to generate network challenges specifications that changes every time the student attempts then. The challenge can then be solved using an emulation package (Dynamips) while a bolt-on package (GNS3) is used to provide the graphical user interface. Once the student has finished a challenge, it will then be graded and feedback presented indicating what was correct and incorrect.
The thesis includes a novel assessment method in grading the academic level of these challenges, based on key academic cogitative levels, such as with Knowledge, Comprehension, Application, Analysis, Synthesis, and Evaluation. There are three example challenges, which are assessed using Bloom’s taxonomy. The challenges include: basic router configuration; EIGRP configuration between two routers; and a redistribution network with a three routers system running EIGRP and OSPF. The thesis outlines the coverage of most of the Bloom layer, apart from the synthesis and evaluation levels, which are covered in the conclusions.
From the results of a questionnaire, two of the most positive aspects of using the framework was that a fully feature IOS command line interface was available for the students to use, and also once they had a mastered a skill they did not have to start from scratch in subsequent exercises,thus not reusing skills that had already mastered. However one of the negative aspects noticed from the
questionnaire was the number of complex steps that was required to be followed to setup the challenge.
The thesis shows the novel results that MAC OS with a Windows XP virtual image produces an average 51% CPU footprint when running the most advanced challenge network topology using the BGP routing protocol, whereas while using the Windows XP and Vista the processor impact is only 1.5-2.5%. It also shows that there is little impact on the disk and memory components. Further investigation was carried out to determine how the Windows XP virtual image would cope with additional routers. It was proven that the processor utilisation was nearly linear, approximately 7% increase per router, going from 2 to 3 to 4, but adding 5 and 6 routers showed that the increase was no longer linear as might have been expected.