Master Thesis Defense: Ting Wang

Speaker: Ting Wang

Supervisor: Dr. B. Jaumard

Examining Committee:
Drs. T. Fevens, L. Narayanan, E. J. Doedel (Chair)

Title:  Time-varying Resilient Virtual Networking Mapping for Multi-location Cloud Data Centers

Date: Wednesday, February 10, 2016

Time: 10:00am

Place: EV 3.309

ABSTRACT

In the currently dominant cloud computing paradigm, applications are being served in data centers (DCs), which are connected to high capacity optical networks. For bandwidth and consequently cost efficiency reasons, in both DC and optical network domains, virtualization of the physical hardware is exploited. In a DC, it means that multiple so-called virtual machines (VMs) are being hosted on the same physical server. Similarly, the network is partitioned into separate virtual networks, thus providing isolation between distinct virtual network operators (VNOs). Thus, the problem of virtual network mapping arises: how to decide which physical resources to allocate for a particular virtual network? In this thesis, we study that problem in the context of cloud computing with multiple DC sites. This introduces additional flexibility, due to the anycast routing principle: we have the freedom to decide at what particular DC location to serve a particular application. We can exploit this choice to minimize the required resources when solving the virtual network mapping problem.

This thesis solves a resilient virtual network mapping problem that optimally decides on the mapping of both network and data center resources, considering time-varying traffic conditions and protecting against possible failures of both network and DC resources. We consider the so-called VNO resilience scheme: rerouting under failure conditions is provided in the virtual network layer. To minimize physical resource capacity requirements, we allow reuse of both network and DC resources: we can reuse the same resources for the rerouting under failure scenarios that are assumed not to occur simultaneously. Since we also protect against DC failures, we allocate backup DC resources, and account for synchronization between primary and backup DCs. To deal with the time variations in the volume and geographical pattern of the application traffic, we investigate the potential benefits (in terms of overall bandwidth requirements) of reconfiguring the virtual network mapping from one time period to the next. We provide models with good scalability, and investigate different scenarios to check whether it is worth to change routing for service requirement between time periods. The results come up with our experiments show that the benefits for rerouting is very limited.