Ammar H. Alhusaini

A unified resource scheduling framework for heterogeneous computing environments

A major challenge in metacomputing systems (computational grids) is to effectively use their shared resources, such as compute cycles, memory, communication network, and data repositories, to optimize desired global objectives. We develop a unified framework for resource scheduling in metacomputing systems where tasks with various requirements are submitted from participant sites. Our goal is to minimize the overall execution time of a collection of application tasks. In our model, each application task is represented by a directed acyclic graph (DAG). A task consists of several subtasks and the resource requirements are specified at subtask level. Our framework is general and it accommodates emerging notions of quality of service (QoS) and advance resource reservations. We present several scheduling algorithms which consider compute resources and data repositories that have advance reservations. As shown by our simulation results, it is advantageous to schedule the system resource separately. Our algorithms have at least 30% improvement over the separated approach with respect to completion time.

An overview of MSHN: the Management System for Heterogeneous Networks

The Management System for Heterogeneous Networks (MSHN) is a resource management system for use in heterogeneous environments. This paper describes the goals of MSHN, its architecture, and both completed and ongoing research experiments. MSHNs main goal is to determine the best way to support the execution of many different applications, each with its own quality of service (QoS) requirements, in a distributed, heterogeneous environment. MSHNs architecture consists of seven distributed, potentially replicated components that communicate with one another using CORBA (Common Object Request Broker Architecture). MSHNs experimental investigations include: the accurate, transparent determination of the end-to-end status of resources; the identification of optimization criteria and how non-determinism and the granularity of models affect the performance of various scheduling heuristics that optimize those criteria; the determination of how security should be incorporated between components as well as how to account for security as a QoS attribute; and the identification of problems inherent in application and system characterization.

A framework for mapping with resource co-allocation in heterogeneous computing systems

In heterogeneous computing systems, an application often requires multiple resources of different types to be allocated simultaneously. This is the resource co-allocation problem. We develop a framework for mapping a collection of applications with resource co-allocation requirements. In our framework, application tasks have two types of constraints to be satisfied: precedence constraints and resource-sharing constraints. We use a graph theoretic framework to capture these constraints. A directed acyclic graph is used to represent precedence constraints of tasks within an application and a compatibility graph is used to represent resource-sharing constraints among tasks of applications. Both these graphs are used to find maximal independent sets of tasks that can be executed concurrently. The objective of the mapping is to minimize the overall schedule length for a given set of applications. We develop heuristic algorithms to solve the mapping problem with resource co-allocation constraints. We also provide a two-phase algorithm that can be used for run-time adaptation. We conducted simulation experiments to evaluate the performance of our heuristic algorithms. Simulation results for our algorithms show a performance improvement of 10% to 30% over a baseline algorithm of list scheduling which considers only the precedence constraints and allocates tasks from the resulting order. This paper demonstrates the importance of considering the co-allocation requirements when mapping applications in heterogeneous computing environments including grid environments.

Run-time adaptation for grid environments

In this paper, we study a general mapping problem where a set of independent tasks compete for the shared resources of a Grid environment. Tasks have resource co-allocationrequirements. Each task requires multiple and different resources to be allocated simultaneously. At run-time, a task may release its allocated resources during its execution and before its completion time. Our objective is to minimize the overall schedule length of all submitted tasks while satisfying all resource sharing constraints among them. We develop a two-phase mapping approach for solving this problem. The first phase of our approach is off-line planning phase where a schedule plan, which gives a scheduling order and resource assignments of tasks, is generated at compile-time. The second phase is run-time adaptationphase. The goal of the second phase is to improve the performance of the schedule plan by adapting to run-time changes such as the early release of resources and the variationin computation and communication costs. Adaptation may involve changing the scheduling order and resource assignments of the original schedule plan. Our experimental results demonstrate the effectiveness of our approach compared to a baseline algorithm that performs no adaptation at run-time and to a dynamic algorithm that performs no planning at compile-time. Our two-phase mapping approach outperforms both algorithms by up to 20% with respect to the overall schedule length.

Adaptive routing and spectrum allocation in elastic optical networks

Elastic optical networks (EONs) have emerged as the preferred technology for future optical networks because of its ability to manage network resources efficiently and provide better spectrum utilization to cope with recent rapid changes in traffic behavior and the tremendous growth in bandwidth demand. An important factor in the success of EONs is routing and spectrum allocation (RSA). In this paper, we propose an adaptive RSA algorithm for EONs. The algorithm relies on the history of carried calls to learn the near-optimal searching sequence of the optical spectrum. The source node measures the gain (cost) of carrying a call at a set of contiguous subcarriers on subsequent (future) call arrival. This gain (cost) will be used to sort the routing and the starting frequencies. We present simulation results to show the efficiency of our new algorithm. Simulation shows savings for the normalized revenue loss can reach up to 70% over the static first-fit RSA with a faster setup time.

An integer programming approach for static mapping of paths onto heterogeneous real-time systems

In this paper, we study the problem of mapping a set of independent paths onto a heterogeneous real-time system. Each path is a pipeline of several stages where each stage consists of one application. Each path has latency and throughput requirements that have to be satisfied. The general goal of this problem is to find an initial static mapping of all applications onto available system resources to maximize the allowable increase in input load, until dynamic remapping is required to avoid Quality of Service (QoS) violations on any path. The objective function considered in this paper is to find a static mapping such that the sum of the latencies of all paths is minimized, while satisfying all latency and throughput requirements. The motivation behind it is to give more room for each path to accommodate additional load at run-time. We develop an integer programming (IP) formulation for modeling this mapping problem. By solving the formulation, a feasible mapping can be found that satisfies latency and throughput requirements of all paths and optimizes the objective function. Two heuristics are given along with the IP formulation to provide efficient solutions. Experimental results show the correctness of the IP formulation and the effectiveness of our heuristics.

Power-efficient routing for SDN with discrete link rates and size-limited flow tables: A tree-based particle swarm optimization approach: A pso-based and power-efficient routing (PSOPR) algorithm

Software-defined networking is a promising networking paradigm for achieving programmability and centralized control in communication networks. These features simplify network management and enable innovation in network applications and services such as routing, virtual machine migration, load balancing, security, access control, and traffic engineering. The routing application can be optimized for power efficiency by routing flows and coalescing them such that the least number of links is activated with the lowest link rates. However, in practice, flow coalescing can generally overflow the flow tables, which are implemented in a size-limited and power-hungry ternary content addressable memory (TCAM). In this paper, a set of practical constraints is imposed to the software-defined networking routing problem, namely, size-limited flow table and discrete link rate constraints, to ensure applicability in real networks. Because the problem is NP-hard and difficult to approximate, a low-complexity particle swarm optimization–based and power-efficient routing (PSOPR) heuristic is proposed. Performance evaluation results revealed that PSOPR achieves more than 90% of the optimal network power consumption while requiring only 0.0045% to 0.9% of the optimal computation time in real-network topologies. In addition, PSOPR generates shorter routes than the optimal routes generated by CPLEX.