{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"institution":[{"id":[{"id":"https:\/\/ror.org\/03mb6wj31","id-type":"ROR","asserted-by":"publisher"},{"id":"https:\/\/www.isni.org\/000000041937028X","id-type":"ISNI","asserted-by":"publisher"},{"id":"https:\/\/www.wikidata.org\/entity\/Q1640731","id-type":"wikidata","asserted-by":"publisher"}],"name":"Universitat Polit\u00e8cnica de Catalunya","acronym":["UPC"]}],"indexed":{"date-parts":[[2026,1,18]],"date-time":"2026-01-18T20:03:40Z","timestamp":1768766620556,"version":"3.49.0"},"reference-count":0,"publisher":"Universitat Polit\u00e8cnica de Catalunya","license":[{"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by-nc-nd\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"abstract":"<jats:p>The simplicity of Internet design has led to enormous growth and innovation. In recent decades several network technologies, services and applications have appeared, which demand specific network requirements for their correct operation.\r\nIn traditional networks, operators are responsible for providing a network configuration sufficiently robust to deal with a wide range of network events and applications. To achieve this is incredibly difficult because: i) the state of the networks can change continuously and today's networks do not provide a mechanism to automatically respond to the wide range of events that may occur and ii) the static nature of current network devices does not permit detailed control-layer configuration, given that the hardware and software are provided by the manufacturer and can not be customized.\r\nThis is the basis of the current, present-day Internet and its architecture, that has grown in an evolutionary fashion from experimental beginnings, rather than from a deliberate strategy. The unpredictable network growth in terms of size and heterogeneity, has exposed a number of fundamental complexities in the current architecture. For instance,  the manual configuration of control functions on network devices that may lead to misconfigurations.\r\nThis is evident that network management requires more intelligent and efficient management systems to coordinate thousands of network elements and applications, the high demand on network performance and growing configuration complexity. In recent decades, several approaches have been introduced in order to improve the network management, such as: MPLS, virtualization and programmable networks. These latter networks have been proposed as a way of facilitating network evolution. In particular, Software Defined Networking (SDN), a networking paradigm focused on allowing software developers to rely on network resources in an easy manner, unifying the state network distribution and a general-purpose technique to manage any type of network in an transparent manner.\r\nIn SDN, network intelligence is logically centralized in software-based controllers (the control layer), and network devices become simple packet forwarding devices (the data layer) that can be programmed via an open interface. By decoupling the control and data layers, network devices can be easily programmed and reconfigured, allowing the behaviour of different types of network devices to be unified. Even though SDN is quite recent, it has already been standardized and implemented in the Internet by several recognized companies such as Google. Several SDN architectures have been proposed to handle current and future network services. However, there are still important research challenges to be addressed in SDN. Some of these current challenges are related to: i) SDN scalability as control is centralized, ii) control layer robustness as any failure can lead to switches to be disconnected from the controller, iii) consistency of network information as wrong decisions can be made affecting network performance and iv) security as controllers can be attacked.\r\nThe purpose of this thesis is to address the  first three of the aforementioned problems. They are addressed from the first premise, ignoring existing approaches offered in traditional networks to remedy some of these issues. First, a controller placement protocol is proposed, taking into account the network\/service requirements. To measure the robustness of a control layer, a robustess metric is designed and evaluated. This metric can also be used to select controller placements in a SDN network that minimize the data loss. Finally, a resource discovery protocol is designed, implemented and evaluated. This protocol discovers any network topology in time efficient, avoiding making assumptions about the network state as it happens in traditional networks.<\/jats:p>\n                <jats:p>En las redes tradicionales, los operadores de red son responsables de proporcionar una configuraci\u00f3n de red lo suficientemente robusta que permita gestionar los diferentes tipos de eventos que puedan afectar el funcionamiento de esta y los requerimientos de los servicios. Esto es dif\u00edcil de alcanzar dado que: i) el funcionamiento de las redes puede variar en cualquier momento y las redes actuales no cuentan con un mecanismo que les permita reaccionar eficientemente al amplio rango de eventos que pueden ocurrir y ii) la naturaleza est\u00e1tica de las elementos de red no permite una detallada configuraci\u00f3n dado que su hardware\/software no pueden ser modificados de una manera eficiente. El impredecible crecimiento de la red en terminos de su tama\u00f1o y su heterogeneidad, han expuesto un n\u00famero de complejidades en la actual arquitectura de red. Primero, los elementos de red tienen que soportar un gran n\u00famero de comandos\/configuraciones sobre un especifico sistema operativo, dificultando la instalaci\u00f3n de un nuevo software sobre ellos, debido a incompatibilidades con el hardware o debido a que el software es incapaz de gestionar las capacidades del hardware. Segundo, la configuraci\u00f3n manual de las funciones de control sobre los elementos de red pueden llevar a configurar err\u00f3neamente las tablas de enrutamiento. Finalmente, la integraci\u00f3n vertical de los middleboxes dificulta a los operadores especificar las pol\u00edticas de alto nivel sobre las tradicionales tecnolog\u00edas de red. La gesti\u00f3n de la red requiere un sistema inteligente y eficiente que coordine: i) los miles de elementos y aplicaciones presentes en la red, ii) la alta demanda sobre el rendimiento de la red y iii) la creciente complejidad en la configuraci\u00f3n de las redes. En las \u00faltimas d\u00e9cadas, diferentes soluciones han sido propuestas con el objetivo de mejorar la gesti\u00f3n de la red, tales como MPLS, virtualizaci\u00f3n y las redes programables. En este \u00faltimo caso, las redes definidas por software o SDNs permiten a los desarrolladores de software gestionar los recursos de red en una manera f\u00e1cil, dado que la distribuci\u00f3n del estado de la red es unificado, lo cual permite gestionar cualquier tipo de red en una manera transparente y en tiempo eficiente. En SDN, la inteligencia de la red esta l\u00f3gicamente centralizada en unos elementos de red llamados controladores, de modo que los dem\u00e1s elementos que act\u00faan en la red solo transmiten paquetes hacia el destino. Estos elementos, son configurados por los controladores a trav\u00e9s de una interface abierta. Es decir, SDN desacopla la capa de control de la capa de datos permitiendo que los elementos de red puedan ser programados y re-configurados independiente del tipo de red. A\u00fan cuando SDN es reciente, este ha sido estandarizado e implementado por diferentes compa\u00f1\u00edas (ej. Google). Sin embargo, hay varios desafios por resolver en SDN a\u00fan. Algunos de estos desafios est\u00e1n relacionados con: i) la escalabilidad de los controladores, como estos est\u00e1n centralizados, ii) la robustez de la capa de control, dado que un fallo en esta puede dejar los elementos de red sin conexi\u00f3n con el controlador, iii) la consistencia de la informaci\u00f3n de control, para evitar tomar decisiones que afecten la operaci\u00f3n de la red, y finalmente iv) la seguridad. En esta tesis, los primeros tres desafios son tratados desde el punto de vista de la localizaci\u00f3n de los controladores en la red, los cuales son seleccionados teniendo en cuenta los requerimientos de los servicios\/aplicaciones y las caracter\u00edsticas de la red. La primera contribuci\u00f3n de esta tesis es un algoritmo que selecciona el n\u00famero de controladores y su localizaci\u00f3n en la red. Un par\u00e1metro de robustez que permite seleccionar los controladores desde los cuales se construye una capa de control robusta y tambi\u00e9n puede medir la robustez de cualquier capa de control, es definida. Finalmente, un protocolo que descubre la topolog\u00eda y caracter\u00edsticas de cualquier red es propuesto y evaluado.<\/jats:p>","DOI":"10.5821\/dissertation-2117-96601","type":"dissertation","created":{"date-parts":[[2023,9,16]],"date-time":"2023-09-16T01:35:11Z","timestamp":1694828111000},"approved":{"date-parts":[[2016,7,6]]},"source":"Crossref","is-referenced-by-count":0,"title":["Scalability and robustness in software-defined networking (SDN)"],"prefix":"10.5821","author":[{"sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Yury Andrea","family":"Jim\u00e9nez Agudelo","sequence":"first","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"3865","container-title":[],"original-title":[],"deposited":{"date-parts":[[2026,1,18]],"date-time":"2026-01-18T06:31:42Z","timestamp":1768717902000},"score":1,"resource":{"primary":{"URL":"https:\/\/hdl.handle.net\/2117\/96601"}},"subtitle":[],"editor":[{"given":"Cristina","family":"Cervell\u00f3 Pastor","sequence":"first","affiliation":[],"role":[{"role":"editor","vocabulary":"crossref"}]}],"short-title":[],"issued":{"date-parts":[[null]]},"references-count":0,"URL":"https:\/\/doi.org\/10.5821\/dissertation-2117-96601","relation":{},"subject":[]}}