Skip to Main Content
The future Internet is driven by applications that require simultaneous access to multiple heterogeneous IT resources interconnected by high-speed optical network. Current research on network and IT resource co-allocation address advanced reservations in which each resource-type is optimised independently. However, emerging applications demand a resource co-allocation model that views and optimizes multiple heterogeneous resources in a holistic manner. A common platform such as the service plane facilitates such a holistically optimised resource allocation model. In this paper, we propose a new efficient online algorithm that views heterogeneous resource-types in a holistic manner. The proposed service-oriented resource allocation model is based on optimised co-scheduling of heterogeneous resources owned by different infrastructure providers. The optimised co-scheduling algorithm aims to serve a higher number of end-users while balancing the load over the multiple resource-types. For this purpose, we introduce a weighted scheduling cost that relies on the simultaneous identification of available resources of each resource-type. The proposed model aims to optimise the efficiency and performance from both the end-users' and infrastructure providers' point of view. A mathematical formulation in the form of an Integer Linear Programming (ILP) is presented to benchmark the performance of the algorithm. Simulations are run to confirm the scalability of the algorithm. The algorithm is also compared against existing resource allocation schemes.
Date of Conference: June 27 2010-July 1 2010