Design and evaluation of virtual network migration mechanisms on shared substrate

Abstract

The Internet faces well-known challenges in realizing modifications to the core architecture. To help overcome these limitations, the use of network virtualization has been proposed. Network virtualization enables the deployment of novel network architectures and services on existing Internet infrastructure. Virtual networks run over physical networks and use Internet paths and protocols as essentially a link layer in the virtual network. Virtual networks can also share the resources in the physical substrate. Effective use of the underlying substrate network requires intelligent placement of virtual networks so that underlying resources do not incur over-subscription. Because virtual networks can come and go over time, and underlying networks can experience their own dynamic changes, virtual networks need to be migrated---re-mapped to the physical network during active operation---to maintain good performance. While virtual network placement, and to a lesser extent migration, has been studied in the past, little attention has been devoted to designing, deploying, and evaluating migration mechanisms for virtual networks. In this dissertation, we design virtual network migration mechanisms for different substrate platforms and further design a system to mitigate the effects of virtual network migration. In particular this dissertation makes the following contributions:

1. With the goal of minimizing the disruption during a virtual network migration, we design three algorithms for scheduling the sequence of virtual router moves that takes a virtual network from its original placement to its new placement.

2. We design and implement a controller-based architecture for virtual network migration on PlanetLab. This work explores the challenges in implementing virtual network migration on real infrastructure. Recommendations are given for infrastructure that support virtual network migration.

3. We propose and implement a mechanism to mitigate the performance degradation resulting from virtual network migration through transport and application layer collaboration. We utilize a centralized controller to notify the end-systems or the gateways about the time of the virtual network migration such that we prevent packet loss to the application traffic of the end-systems.