In the last few years, the development of 5G networks has brought countless benefits for network operators including greater flexibility, greater scalability and a more efficient resources utilization. In addition, it has offered greater network speed to end users, allowing better use of services and applications. In this context, Network Slicing constitutes one of the key aspects of the development of modern networks, allowing the same network infrastructure to host numerous services, characterized by different Quality of Service (QoS). These services are managed by Service Providers and are represented as virtual networks. On the other hand, network resources are managed by an Infrastructure Provider, which assigns them to the various Service Providers. Network slicing provides greater flexibility to the network, meeting the the most different service requirements in terms of latency, traffic volume density, experienced end-user throughput, energy consumption and reliability. It also allows the separation of network resources assigned to different virtual networks, thus guaranteeing the security of each virtual network. However, in addition to bringing advantages, network slicing incorporates numerous challenges in the management of network resources and services. Among them, network reliability has become one of the most important aspects, which 5G networks must take into account. For this reason, in this thesis we focus on the Survivable Virtual Network Mapping problem (SVNM), which consists of assigning physical network resources to the virtual networks supporting the offered services, in such a way as to guarantee the survivability of network services against physical link failures. To solve this problem, we propose two new techniques that provide an SVNM allowing the capacity sharing between different virtual networks and the use of a shared spare slice. Capacity sharing is achieved through the common nodes of virtual networks, which we define as inter-slice gateways and is appropriately limited to ensure the isolation of virtual networks as much as possible. To evaluate the benefits of the proposed techniques we identify seven network scenarios in which the two strategies are combined differently. We also develop four Integer Linear Programming (ILP) models and, to cope with scalability problems, we propose a heuristic algorithm that adapts to the network scenarios considered. The proposed techniques are applicable to any type of physical or virtual network considered. Results show that the two proposed techniques bring advantages over the cases of SVNM against single link failures and SVNM against double link failures, achieving the survivability of virtual networks against double link failures with lower network costs. These advantages increase as the number of virtual networks considered in the problem increases as the two techniques depend on the common nodes of the virtual networks and a greater number of virtual networks corresponds to a greater possibility of sharing capacity.
Negli ultimi anni, lo sviluppo delle reti 5G ha portato innumerevoli vantaggi per gli operatori di rete tra cui una maggiore flessibilità, una maggiore scalabilità e un utilizzo più efficiente delle risorse. Inoltre, ha offerto una maggiore velocità di rete agli utenti finali, permettendo una migliore fruizione dei servizi e delle applicazioni. In questo contesto, il Network Slicing costituisce uno degli aspetti chiave dello sviluppo delle reti moderne, permettendo ad una stessa infrastruttura di rete di ospitare numerosi servizi, caratterizzati da Quality of Service (QoS) differenti. Questi servizi vengono gestiti da Service Providers e vengono rappresentati come reti virtuali. Le risorse di rete invece, sono gestite da un Infrastructure Provider che provvede ad assegnarle ai diversi Service Providers. Il network slicing fornisce una maggiore flessibilità alla rete, soddisfando i più diversi requisiti di servizio in termini di latenza, densità di volume di traffico, throughput dell'utente finale, consumo di energia e affidabilità. Consente inoltre la separazione delle risorse di rete assegnate a diverse reti virtuali, garantendo così la sicurezza di ogni rete virtuale. Il network slicing però, oltre a portare dei vantaggi, ha portato numerose sfide nella gestione delle risorse e dei servizi di rete. Tra questi, l'affidabilità della rete è diventata uno degli aspetti più importanti, di cui le reti 5G devono tenere conto. Per questo motivo, in questa tesi ci concentriamo sul problema del Survivable Virtual Network Mapping (SVNM), che consiste nell'assegnare risorse fisiche di rete alle reti virtuali supportando i servizi offerti, in modo tale da garantire la sopravvivenza dei servizi di rete contro i guasti di link fisici. Per risolvere questo problema, proponiamo due nuove tecniche che forniscono un SVNM permettendo la condivisione di capacità tra le diverse reti virtuali e l'utilizzo di una spare slice condivisa. La condivisione di capacità viene realizzata tramite i nodi comuni alle network slices, da noi definiti come inter-slice gateways e viene opportunamente limitata per garantire il più possibile l'isolamento delle reti virtuali. Per valutare i vantaggi delle tecniche proposte identifichiamo sette scenari di rete in cui le due strategie vengono combinate in maniera differente. Sviluppiamo inoltre quattro modelli di Integer Linear Programming (ILP) e, per far fronte ai problemi di scalabilità, proponiamo un algoritmo euristico che si adatta agli scenari di rete considerati. Le due tecniche proposte sono applicabili a qualsiasi tipo di rete fisica o virtuale considerata. I risultati mostrano che le due tecniche proposte portano dei vantaggi rispetto ai casi di SVNM contro i fallimenti singoli di link e di SVNM contro i fallimenti doppi di link, raggiungendo la sopravvivenza delle reti virtuali contro i guasti doppi di link con dei costi di rete inferiori. Questi vantaggi aumentano all'aumentare del numero di reti virtuali considerate nel problema in quanto le due tecniche dipendono dai nodi comuni alle reti virtuali e a un maggior numero di network slices corrisponde una maggiore possibilità di condivisione di capacità.
New techniques for survivable virtual network mapping based on capacity sharing against double link failures
VIADANA, EMANUELE
2019/2020
Abstract
In the last few years, the development of 5G networks has brought countless benefits for network operators including greater flexibility, greater scalability and a more efficient resources utilization. In addition, it has offered greater network speed to end users, allowing better use of services and applications. In this context, Network Slicing constitutes one of the key aspects of the development of modern networks, allowing the same network infrastructure to host numerous services, characterized by different Quality of Service (QoS). These services are managed by Service Providers and are represented as virtual networks. On the other hand, network resources are managed by an Infrastructure Provider, which assigns them to the various Service Providers. Network slicing provides greater flexibility to the network, meeting the the most different service requirements in terms of latency, traffic volume density, experienced end-user throughput, energy consumption and reliability. It also allows the separation of network resources assigned to different virtual networks, thus guaranteeing the security of each virtual network. However, in addition to bringing advantages, network slicing incorporates numerous challenges in the management of network resources and services. Among them, network reliability has become one of the most important aspects, which 5G networks must take into account. For this reason, in this thesis we focus on the Survivable Virtual Network Mapping problem (SVNM), which consists of assigning physical network resources to the virtual networks supporting the offered services, in such a way as to guarantee the survivability of network services against physical link failures. To solve this problem, we propose two new techniques that provide an SVNM allowing the capacity sharing between different virtual networks and the use of a shared spare slice. Capacity sharing is achieved through the common nodes of virtual networks, which we define as inter-slice gateways and is appropriately limited to ensure the isolation of virtual networks as much as possible. To evaluate the benefits of the proposed techniques we identify seven network scenarios in which the two strategies are combined differently. We also develop four Integer Linear Programming (ILP) models and, to cope with scalability problems, we propose a heuristic algorithm that adapts to the network scenarios considered. The proposed techniques are applicable to any type of physical or virtual network considered. Results show that the two proposed techniques bring advantages over the cases of SVNM against single link failures and SVNM against double link failures, achieving the survivability of virtual networks against double link failures with lower network costs. These advantages increase as the number of virtual networks considered in the problem increases as the two techniques depend on the common nodes of the virtual networks and a greater number of virtual networks corresponds to a greater possibility of sharing capacity.File | Dimensione | Formato | |
---|---|---|---|
2021_04_Viadana.pdf
accessibile in internet per tutti
Descrizione: Testo della tesi
Dimensione
2.08 MB
Formato
Adobe PDF
|
2.08 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/175743