Abstract:

Computer network design and operation are very complex tasks due to the distributed nature of its operation, the high number of configuration parameters, the different possible failures they may face, etc. A theoretical exclusive learning of this technology is clearly not enough for a future engineer, but experimentation with real equipment requires lots of resources, and therefore it is expensive in terms of equipment cost and lecturer's time. Aware of the need for real experimentation to form the new generation of Computer Science and Telematics engineers, the department of Telematics Engineering of the University Carlos III of Madrid have done a great effort by building a comprehensive set of networking laboratories that merge virtual and real equipment to provide the student with a playground to train its theoretical knowledge in real life scenarios and applications. This laboratory is composed of several assignments aimed at letting the student configure and test real protocols, such as RIP or BGP, with large-scale deployments, during the always limited time available in the classes. In order to extend this limited time, we have developed a virtual environment with virtualized versions of the hardware used in the classes, through which the students are able to replicate the same experiments performed during the physical laboratories.

The virtual and real laboratories experiences have been already in place for some years now, and although the acceptance by the students has been very satisfactory, there are some drawbacks we have been identifying during the last years:

• We have detected that rookie students do not take fully advantage of the laboratory, requiring several sessions using the same equipment to develop the capacities of understanding what is happening in the testbed, how to detect problems and how to validate if the configuration applied works.
• The consequence of the students not being able to identify issues in the testbed makes the lecturer spend a long time per lab-group, performing the required methodology to find what is not working and why.
• In a class where several groups are doing the same lab assignment, the fact that the lecturer has to spend a long time per group reduces the time the students are actually playing with the configuration and the testbed, since once they detect some issue they usually stop and wait until the lecturer reaches their group and helps them solve the problem.
• The networking labs integrated in some lectures require an individual exam where the students show their recently acquired knowledge about network design, configuration and operation. Due to the amount of time required to validate, find errors and let the student try to solve them, these exams take an extremely long time to complete if performed using real equipment in a sequential manner.

In order to solve all of these issues, in this paper we present a system for automatic correction and validation of real and virtualized networking laboratories that we have been developing during the last year. This system enables the students to validate if a given setup works as expected by applying a battery of tests defined by the lecturer, in an automatic way, providing feedback about the issues found and hence enabling the student to identify error and miss-configurations in an autonomous way.

Keywords:

Networking Labs, grading and correction, automatic tools.