Project Details
Description
Virtualized infrastructure is becoming increasingly routine, and so the ability to manage it well is important. Cloud computing is built on virtualized computing infrastructure. Software Defined Networks are virtualized data networks. The combination of virtualized computing and communications constitutes the next wave of innovation in provisioning bespoke infrastructure. This project addresses ways of making such virtualized infrastructure more reliable, by planning and managing the migration of the virtual network structure and associated computation from one set of supporting physical infrastruture to another. Triggers for VN migration include attacks, massive failures, performance collapse or degradation, or changes in the service demands that require an increase or decrease in resource allocation in order to both satisfy users and reduce power consumption. This ability to adapt to network dynamics is referred to as VI migration (VIM).
There are many challenges in deploying VIM for the applications above. First, different network agility objectives might require different VIM models and techniques. For example, the trigger to migrate for proactive defense to defeat reconnaissance would need to be time based , whereas, for power management the trigger would be significant change of network traffic load. Second, VN migrations may need to be planned in advance (for example, in the case of proactive defense) but in some cases they cannot be planned in advance and need to be deployed in real time (in response to faults, for example). Third, although most existing theories permit reasoning about correctness of network operation only once as the system stabilizes to a static state, investigating formal frameworks that support migration strategies that are correct-by-construction under continuous movement is a prime research task. To address these challenges this project will develop a VIM architecture that is designed to work in the context of the GENI and PlanetLab virtualization technologies, but which could be used as the basis for other industrial systems. The architecture consists of two main components: a Strategy Synthesizer block and a Migration Mechanism block. The inputs to the Strategy Synthesizer block uses models of the VI and the Migration Mechanism block interacts directly with the substrate to implement the desired migration.
VIM is a powerful technique for energy control in Cloud infrastructure and effective tools for managing VIM are expected to facilitate significant savings in IT energy consumption. In addition to the potential commercial and network security impacts of the work, the PIs have identified local and outreach activities that further the participation of women and minorities. The experimental and evaluation aspects of the work will be incorporated as project elements in graduate courses.
Status | Finished |
---|---|
Effective start/end date | 1/9/13 → 31/8/18 |
Links | https://www.nsf.gov/awardsearch/showAward?AWD_ID=1320662 |
Funding
- National Science Foundation: US$249,999.00
ASJC Scopus Subject Areas
- Computer Networks and Communications