In this thesis we consider the problem of providing QoS guarantees to Grid users through advance reservation. Advance reservation mechanisms provide the ability to allocate resources to users based on agreed-upon QoS requirements and increase the predictability of a Grid system, yet incorporating such mechanisms into current Grid environments has proven to be a challenging task due to the resulting resource fragmentation. In view of these observations we have devised efficient scheduling algorithms that support advance reservations. We can organize this thesis in two parts.;We first use concepts from computational geometry and efficient data structures to present a framework for tackling the resource fragmentation, and for formulating a suite of scheduling strategies. We also develop efficient implementations of the scheduling algorithms that scale to large Grids. We conduct a comprehensive performance evaluation study using simulation, and we present numerical results to demonstrate that our algorithms perform well across several metrics that reflect both user- and system-specific goals.;Advance reservations has also been proposed as one mechanisms to provide Grid resource managers with the ability to co-allocate resources. Co-allocation of resources is one problem that has gained increasing attention due to the emergence of complex applications that require orchestration of resources never envisioned before. In practice, a co-allocation request can be handled manually as a set of individual advance reservations requests. However, such a solution can be computationally expensive and inappropriate for time-sensitive applications. Furthermore, the trend towards more dynamic solutions has emphasized the need for more automatic solutions. As a second contribution, in this thesis we design and develop a co-allocation algorithm that is efficient in co-allocating resources while respecting the atomicity of the co-allocation request and improving user and system performance. This is achieved by quantizing the temporal space and using efficient 2-dimensional balanced search trees. We perform a comparative analysis of our algorithm by means of simulations driven by workloads from real systems and show that our algorithm scales to Grid systems with large number of resources and heavy workloads.
展开▼
