New services emerging in metro networks force network operators to seek new approaches for provisioning of the resources required by these services in the network. The traditional approach that leverages the single-purpose hardware devices imposes high Capital Expenditure (CapEx) and Operational Expenditure (OpEx) on network operators and will not provide an agile way to provision services. In other words, the hardware devices used as middle-boxes only provide a single functionality, therefore, to provide multiple functionalities in the network, several middleboxes are needed, resulting in high CapEx and the OpEx, especially in terms of energy cost. Furthermore, these devices are typically expensive and have limited life cycle. Network Function Virtualization as a new paradigm shift has emerged to enable network operators reduce the CapEx and OpEx by replacing hardware devices by software instances that can run on generic purpose servers and switches. These software instance that can run on any generic server are referred to as Virtual Network Functions (VNFs). However, the decision about where to place these VNFs in the network and how to concatenate them in the specific order to provision a Service Chain (SC) is not trivial. In this thesis, we mainly focus on the problem of VNF placement in optical metro networks and develop novel algorithms to perform VNF placement for dynamic service chaining. We consider the problem of SC provisioning under different scenarios: i) Latency-aware service chaining for which we propose an algorithm that based on the latency requirements of each SC decides about performing traffic grooming and bandwidth assignment of SCs. ii) Reprovisioning for which we consider the situation in which the placement of VNFs can be changed in the network to achieve higher performance in terms of number of SCs provisioned in the network. iii) Protection for which we propose algorithms to provide different protection strategies for dynamic service chaining. iv) Filterless for which we propose an algorithm to perform dynamic service chaining in filterless optical networks. We further perform techno-economic analysis of different VNF placement strategies to figure out the best trade-off between bandwidth and computational cost in metro networks to guarantee a specific service availability defined in the Service Level Agreement of the users.
I nuovi servizi che emergono nelle reti metropolitane costringono gli operatori di rete a cercare nuovi approcci per fornire delle risorse richieste da questi servizi nella rete. L'approccio tradizionale che sfrutta i dispositivi hardware monouso impone elevate spese in Capital Expenditure (CapEx) e Operational Expenditure (OpEx) agli operatori di rete e non offrirà un modo agile per fornire servizi. In altre parole, i dispositivi hardware utilizzati come middle-box forniscono solo una singola funzionalità, quindi, per fornire più funzionalità nella rete, sono necessari diversi middlebox, con conseguente elevato CapEx e OpEx, soprattutto in termini di costo energetico. Inoltre, questi dispositivi sono tipicamente costosi e hanno un ciclo di vita limitato. La virtualizzazione delle funzioni di rete come nuovo cambiamento di paradigma è emerso per consentire agli operatori di rete di ridurre CapEx e OpEx sostituendo i dispositivi hardware con istanze software che possono essere eseguite su server e switch generici. Queste istanze software che possono essere eseguite su qualsiasi server generico sono denominate Virtual Network Functions (VNFs). Tuttavia, la decisione su dove posizionare questi VNF nella rete e su come concatenarli nell'ordine specifico per fornire un Service Chain (SC) non è banale. In questa tesi, ci concentriamo principalmente sul problema del posizionamento VNF nelle reti metropolitane ottiche e sviluppiamo nuovi algoritmi per eseguire il posizionamento VNF per il Service Chaining dinamico. Consideriamo il problema del fornitura di SC in diversi scenari: i) Concatenamento di servizi con riconoscimento della latenza per il quale proponiamo un algoritmo che in base ai requisiti di latenza di ogni SC decide di eseguire il grooming del traffico e l'assegnazione della larghezza di banda degli SC. ii) Reprovisioning per il quale si considera la situazione in cui il posizionamento delle VNF può essere modificato nella rete per ottenere prestazioni più elevate in termini di numero dei SC forniti nella rete. iii) Protezione per la quale proponiamo algoritmi per fornire diverse strategie di protezione per il concatenamento dinamico dei servizi. iv) Filterless per il quale proponiamo un algoritmo per eseguire il concatenamento dinamico dei servizi in reti ottiche senza filtro. Eseguiamo inoltre analisi tecnico-economiche di diverse strategie di posizionamento VNF per capire il miglior compromesso tra larghezza di banda e costo computazionale nelle reti metropolitane per garantire una specifica disponibilità del servizio definita nel Service Level Agreement degli utenti.
Dynamic management of virtual network functions in optical metro network architectures for 5G communications
Askari, Leila
2020/2021
Abstract
New services emerging in metro networks force network operators to seek new approaches for provisioning of the resources required by these services in the network. The traditional approach that leverages the single-purpose hardware devices imposes high Capital Expenditure (CapEx) and Operational Expenditure (OpEx) on network operators and will not provide an agile way to provision services. In other words, the hardware devices used as middle-boxes only provide a single functionality, therefore, to provide multiple functionalities in the network, several middleboxes are needed, resulting in high CapEx and the OpEx, especially in terms of energy cost. Furthermore, these devices are typically expensive and have limited life cycle. Network Function Virtualization as a new paradigm shift has emerged to enable network operators reduce the CapEx and OpEx by replacing hardware devices by software instances that can run on generic purpose servers and switches. These software instance that can run on any generic server are referred to as Virtual Network Functions (VNFs). However, the decision about where to place these VNFs in the network and how to concatenate them in the specific order to provision a Service Chain (SC) is not trivial. In this thesis, we mainly focus on the problem of VNF placement in optical metro networks and develop novel algorithms to perform VNF placement for dynamic service chaining. We consider the problem of SC provisioning under different scenarios: i) Latency-aware service chaining for which we propose an algorithm that based on the latency requirements of each SC decides about performing traffic grooming and bandwidth assignment of SCs. ii) Reprovisioning for which we consider the situation in which the placement of VNFs can be changed in the network to achieve higher performance in terms of number of SCs provisioned in the network. iii) Protection for which we propose algorithms to provide different protection strategies for dynamic service chaining. iv) Filterless for which we propose an algorithm to perform dynamic service chaining in filterless optical networks. We further perform techno-economic analysis of different VNF placement strategies to figure out the best trade-off between bandwidth and computational cost in metro networks to guarantee a specific service availability defined in the Service Level Agreement of the users.| File | Dimensione | Formato | |
|---|---|---|---|
|
PhD_Thesis.pdf
non accessibile
Dimensione
3.51 MB
Formato
Adobe PDF
|
3.51 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/10589/177109