With the advent of next-generation scientific applications, the workflow-based computing technology has become an indispensable research method for managing and streamlining large-scale distributed data processing. This paper investigates a problem of mapping distributed workflows for streaming data processing in faulty networks where nodes and links are subject to probabilistic failures. We formulate this problem as a bi-objective optimization problem in terms of both throughput and reliability, and propose a decentralized layer-oriented method to achieve high throughput for smooth data flow while satisfying a prespecified overall failure rate bound for a guaranteed level of reliability. The superiority of the proposed mapping solution is illustrated by both extensive simulation-based performance comparisons with existing algorithms and experimental results from a real-life scientific workflow deployed in wide-area networks.

As virtualization becomes more and more popular, how to guarantee survivability of a virtual infrastructure (VI) over a wide-area optical network is increasingly important. In this paper, we approach the problem of survivable VI mapping (SVIM) from a few unique perspectives. One of the most distinguishing perspectives is that a large-scale regional failure could destroy one or more facility nodes to which some VI nodes are mapped. Accordingly, redundant facility nodes at different geographical locations and redundant optical connections have to be provisioned such that the VI can still be mapped after the failure. Another distinguishing perspective is that with failure-dependent protection, the SVIM problem can be decomposed into several instances of the basic non-survivable VI mapping (NSVIM) problem, whose solution permits effective sharing of the redundant resources among all failures. In this paper, we first formulate the minimum-cost SVIM problem using mixed integer linear programming (MILP). We then propose an efficient heuristic solution to NSVIM, based on which two novel heuristic SVIM algorithms called Separate Optimization with Unconstrained Mapping (SOUM) and Incremental Optimization with Constrained Mapping (IOCM). Simulations are performed to study and compare the performance of the MILP and heuristics.

Under the emerging paradigm of federated computing services (FCS), one can create multiple virtual infrastructures (VI), one for each distributed computing application or service offering. Each VI may consist of a number of geographically distributed computing clusters that are connected with a set of dedicated circuits. In general, a VI submitted by a user to the FCS provider is unmapped at the time of the submission in that the user does not specify which computing clusters to use. The primary challenge to the FCS provider in supporting these novel applications is to establish a VI optimally over the substrate network such as a WDM network connecting many computing clusters. In this paper, we study the optimization problems of jointly allocating computing and wavelength resources for establishing Vis. First, we devise a branch and bound algorithm based on the decomposition and Lagrangian relaxation techniques to obtain the exact optimal solution with the objective being the minimization of the resource leasing cost of a VI. Second, we propose efficient heuristics to deal with a large number of online requests for Vis and compare their performance with the optimal solution.

Data-intensive Grid applications require huge data transferring between multiple geographically separated computing nodes where computing tasks are executed. For a future WDM network to efficiently support this type of emerging applications, traditional approaches to establishing lightpaths between given source destination pairs are not sufficient because a computing task may be executed on any one of several computing nodes having the necessary resources. Therefore, lightpath establishment has to be considered jointly with task scheduling to achieve best performance. We study the optimization problems of jointly scheduling both computing resources and network resources. We first present the formulation of two optimization problems with the objectives being the minimization of the completion time of a job and minimization of the resource usage/cost to satisfy a job with a deadline respectively. When the objective is to minimize the completion time, we devise an optimal algorithm for a special type of applications. Furthermore, we propose efficient heuristics to deal with general applications with either optimization objective and demonstrate their good performances via simulation.

In this paper, we present a new integrated optical network architecture called polymorphous, agile and transparent optical networks (PATON) which combines different resource scheduling schemes based on polymorphous optical burst switching (POBS) suitable for a variety of applications and services, and address several key technical issues related to the control and management of PATON. These issues include traffic adaptation and grooming, signaling and control packet format, polymorphic control for resource scheduling, and extentions to GMPLS. The resulting PATON compares favorably with other existing approaches presented in our related work [X. Liu et al., 2006].