Resource Allocation Techniques


Server Selection and Admission Control in IPTV Networks

It is becoming more common for end users to consume video content over their home broadband connections. Service providers are interested in methods of delivering premium quality video over unmanaged Internet backbones to end users while maintaining a satisfactory level of quality. This research focuses on investigating traffic management strategies to maximise the quality of service delivered to premium customers in light of dynamic changes in network resource availability. The strategies employed are that of identifying appropriate servers within the cloud ideally suited to serve content to the end user, depending on the availability of resources between the two points.

VM placement in clouds

We address the problem of maximizing the revenue generated from the placement of sets of virtual machines (VMs) on a set of physical machines (PMs) in an Infrastructure-as-a-Service (IaaS) cloud environment. Requests for placement of sets of VMs can have specified combinations of placement constraints (e.g. full deployment or anti-collocation) and specified minimal resource requirements (e.g., memory or CPU).  We describe two approaches. The first is based on the formulation of an integer linear programming (ILP) problem, the solution to which provides an optimal VM placement. The second approach is a heuristic for this ILP based on classifying the requests into different categories and satisfying the constraints in a particular order using variants of greedy algorithms for the multi-dimensional vector bin packing problem. Given a model of VM placement constraints, offered resources and requests with multiple VM types, both approaches devise a placement plan in a way that maximizes revenue, having due regard both to customer requirements and PM resource profiles. We evaluate the relative performance of the solutions by means of numerical experiments. The results suggest the optimal solution is not practical for medium to large problems. For smaller problems the heuristic solution provides placements close to the optimal solution. Moreover, for larger scale problems our results show that the heuristic is practical in terms of its runtime efficiency, so that it can be used for online placement of arriving requests for placement of VM sets.

Energy Efficient Resource Allocation in Data Centres

In recent years the research community has been actively looking for new solutions to support energy efficient communication networks. One of the growing concerns in this regard is the increasing role played by data centres in energy consumption. As the number of services and content on the Internet grow, we are seeing higher energy consumed, not only for powering the servers but also for cooling. From the global community perspective, there has also been increasing awareness of continued reliance on fossil fuels, which has led to increased global carbon emission. Therefore, one of the aims of our research has been to use intelligent optimisation algorithms to determine optimal resource allocations (e.g. Services) in order to increase usage of renewable energy (hence minimising carbon emissions), while at the same time considering a range of mitigating factors such as energy cost, resource migration distance etc.

Members

Dr. Lei Shi, Dr. Ray Carroll, Dr. Alan Davy, Brian Meskill, Cathal O’Connor

Collaboration

Prof. Jordi Domingo Pascual, Dr. Albert Cabellos, Universitat Politècnica de Catalunya

Projects

(SFI) Fame, (IRC) Maven

Recent Publications

Brian Meskill, Alan Davy, and Brendan Jennings. The Impact of the Complexity of Topologies used in Comparative Analyses of Congestion-based Available Bandwidth Estimation Tools. In Proc. 10th Information Technology & Telecommunications Conference (IT’&T 2010), 2010.

Brian Meskill, Alan Davy, and Brendan Jennings. Server Selection and Admission Control for IP-based Video on Demand Using Available Bandwidth Estimation. In Proc. 36th Annual IEEE Conference on Local Computer Networks (LCN 2011) [short paper], pages 255–258. IEEE, October 2011. (doi:10.1109/LCN.2011.6115202)

Brian Meskill, Alan Davy, and Brendan Jennings. Revenue-Maximizing Server Selection and Admission Control for IPTV Content Servers using Available Bandwidth Estimates. In Proc. 2012 IEEE/IFIP Network Operations and Management Symposium (NOMS 2012), pages 319–326. IEEE, 2012.

Cristian Olariu, John Fitzpatrick, Philip Perry, and Liam Murphy. A QoS based call admission control and resource allocation mechanism for LTE femtocell deployment. In Proc. 2012 IEEE International Conference on Consumer Communications and Networking (CCNC 2012). IEEE, January 2012.

Lei Shi, Bernard Butler, Runxin Wang, Dmitri Botvich, and Brendan Jennings. Optimal placement of virtual machines with different placement constraints in iaas clouds. In Proc. 2012 China Ireland Symposium on ICT and Energy Effciency (CIICT 2012). IET, 2012.

Lei Shi, Bernard Butler, Dmitri Botvich, and Brendan Jennings. Provisioning of Requests for Virtual Machine Sets with Placement Constraints in IaaS Clouds. In Proc. 13th IFIP/IEEE International Conference on Integrated Network Management (IM 2013), to appear. IEEE, 2013.