During the last years it has been emerging a trend at universities, large enterprises, government institutions, hospitals and public institutions towards acquiring and deploying their own dark fibre or wavelength networks as opposed to purchasing bandwidth network services to the traditional carriers.
These institutions usually follow the condominium model to build and deploy their network. In this model several parties get together in a joint effort to purchase the network equipment and deploy the dark fibre.
Each institution gets a subset of the deployed fibre and a subset of the available ports in the network gear proportional to their initial investment. Even though there is a common physical substrate, each institution wants to manage their resources independently of the other organizations that share the physical substrate, and deploy its own routing protocols or protection schemes. Traditional control plane architectures cannot address the requirements of this type of networks, because they assume that a single entity has administrative control of all the network elements in a physical domain.
There is a growing consensus within the Internet community in that the Internet is a huge success, but that it has a series of unsolved issues (such as security or mobility) that the original architecture didn't address. Consequently the Internet architecture needs to be redesigned following either an evolutionary or a clean slate approach. However, deploying and testing new network architectures that can potentially disrupt the current Internet is not possible. At the same time, new architectures have to be validated in environments that are as close as possible to the real Internet environment (simulations and traditional testbeds are not enough to validate a candidate architecture). Several international research efforts such as GENI or FEDERICA, are trying to address the problem by using network virtualization techniques to allow different networks to run in parallel sharing a common physical substrate. This way new Internet architectures can be deployed and tested in the same physical substrate as the current Internet, but they can still be isolated from each other. The requirements introduced by this scenario are quite similar to the condominium network use case: a common physical substrate is partitioned in a set of segments, also called slices, and each slice is controlled by an independent organization.
At the same time a new set of bandwidth intensive applications are emerging. E-science applications, the science that uses cyber infrastructure and high performance networks as basic instruments to perform the research, grid applications and high definition digital media streaming produce such a big amount of data that sometimes justify dedicating a network to a single application. However, in order to efficiently manage the resources, these applications should be able to configure the network in the way it better suits their needs. This is the reason why these applications are often known as network aware applications.
The objective of this thesis is to investigate, design, prototype, test, deploy and validate intelligent software solutions that can address the scenarios identified in the last paragraphs. The first section of the thesis explains the results of the work carried out during the UCLPv1 research program (UCLPv1 for optical networks, UCLPv1 for Ethernet and MPLS networks, UCLP and GridOn integration, discussion of the results). The second section of the thesis presents the research carried out during the UCLPv2 program (UCLPv2, the Resource Management Center concept, UCLP CE, the advance reservation service, discussion of the results). The third section of the thesis deals with the generalization of the UCLPv2 concepts (the IaaS Framework) and its application to optical networks (Argia) and IP Networks (MANTICORE). Finally, the last section of the thesis presents the overall conclusions and future work.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados