École de technologie supérieure
Performance and survivability of service function chains in virtualized environments
Abstract
dc:descriptionNetwork Function Virtualization is an emergent technology enabling the deployment of network services in the form of service function chains in an efficient and flexible manner. A service chain is composed of virtual network function that could implement different functions (e.g., IDS, Firewall or others). Despite their benefits, many challenges are faced in order to provisioning such chains, manage their resources and ensure their service continuity. In this thesis, we address tow key challenges pertaining to the management of service function chains, namely 1) the synchronization among virtual network functions composing the chain, 2) how to ensure the survivability of such service chains in case failures. The first challenge addressed by this thesis is how to ensure the synchronization between multiple instances of the same network function with minimal costs and acceptable delay. Specifically, we propose a technique that identifies the optimal communication pattern between the instances in order to minimize the synchronization costs and delay. We also propose a novel network function named Synchronization Function able to carry out data collection and further minimize these costs. We then formulate the problem of finding the optimal communication pattern, the optimal number and placement of synchronization functions as an Integer Linear Program aiming at minimizing the synchronization costs. We also put forward three greedy algorithms to cope with large-scale scenarios of the problem and explore the use of network function migration to further reduce costs. Extensive simulations show that the proposed algorithms efficiently find near-optimal solutions with minimal computation time and provide better results compared to existing solutions. The second challenge that we have addressed is how to ensure the survivability of service function chains in case of multiple and simultaneous failures. We introduce a survivability management framework to efficiently manage backups for network functions. We mathematically model the problem of service chain survivability as an Integer Linear Program that determines the minimal number of shared backups and their optimal location in the physical infrastructure while ensuring a minimal operational cost. Finally, we propose two greedy algorithms to deal with the problem’s large-scale scenarios and we investigated the use of traffic prediction to further reduce costs. Simulations demonstrate how the proposed algorithms provide near-to-optimal solutions.
Degree
thesis:*- Name thesis:degree_name
- Thèse de doctorat électronique
- Level thesis:degree_level
- phd
- Grantor dc:publisher
- École de technologie supérieure
- Year dc:date
- 2022
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Alomari, Zakaria
Subjects
dc:subject × 1Rights
- Language dc:language
- en
Identifiers
dc:identifier.*- Identifier
- INSERT-YOUR-LIBRARY-ID-HERE3212