CMPSCI 691FF: Algorithmics for Internet-Based Computing

Instructor:	Arnold L. Rosenberg	Office:	CMPS 308
		Email:	rsnbrg@cs.umass.edu
		Office Hours:	By appointment
Class Mtgs:	TuTh, 11:15-12:30	CMPS 140

Prerequisite: CMPSCI 611 plus at least one systems core course with a grade of B or better.

Overview. Advances in technology have rendered the Internet a viable medium for a new genre of ``collaborative'' computing (that we call Internet-based computing [IC, for short]), wherein multiple computing agents, possibly widely dispersed geographically, cooperate in the solution of a single computational problem. In a series of papers [3, 4, 5, 12, 13, 145] we are currently developing an algorithmic framework for scheduling computations having intertask dependencies, for the several modalities of IC--including Grid computing (cf. [1, 6, 7]) global computing (cf. [2]), and Web computing (cf. [9]). This new IC-scheduling theory aims to craft schedules that maximize the rate at which tasks are rendered eligible for allocation to remote clients (hence for execution), with the dual goal of: (a) enhancing the effective utilization of remote workers, by always having a task to allocate to an available worker; (b) lessening the likelihood of the ``gridlock'' that can arise when a computation stalls pending computation of already-allocated tasks. Preliminary simulation experiments reported in [10, 8] bolster our hope that IC-scheduling theory will be the underpinnings of a theory of scheduling complex computations for IC.

The course. This course will be devoted to studying and advancing IC-scheduling theory. The ``studying'' will consist of reading (portions of) several of the papers cited in the preceding paragraph. Each student will be responsible for presenting at least one, and at most two, papers during the semester. The ``advancing'' will consist in a piece of research that advances the theory either via improved/extended algorithms or via an improved/extended simulation study. Each student will participate in such a study, chosen in consultation with the instructor. The studies can be either solo or in teams (at the student's discretion). Part of the excitement will be deciding what kinds of studies are needed/appropriate.

Grades. The grading will be based on the quality of a student's presentation(s) and the perceived effort put into the research topic.

References. The following references are for educational purposes only, not for other distribution.

1. R. Buyya, D. Abramson, J. Giddy (2001): A case for economy Grid architecture for service oriented Grid computing. 10th Heterogeneous Computing Wkshp.

2. W. Cirne and K. Marzullo (1999): The Computational Co-Op: gathering clusters into a metacomputer. 13th Intl. Parallel Processing Symp., 160-166.

3. G. Cordasco, G. Malewicz, A.L. Rosenberg (2006): Advances in a dag-scheduling theory for Internet-based computing (in PDF). Submitted for publication. See also, On scheduling expansive and reductive dags for Internet-based computing. 26th Intl. Conf. on Distributed Computing Systems, 2006.

4. G. Cordasco, G. Malewicz, A.L. Rosenberg (2006): Applying IC-scheduling theory to familiar classes of computations (in PDF). Typescript (draft), Univ. Massachusetts.

5. G. Cordasco, G. Malewicz, A.L. Rosenberg (2006): Extending IC-Scheduling Theory via the Sweep Algorithm (in PDF). Typescript (draft), Univ. Massachusetts.

6. I. Foster and C. Kesselman [eds.] (2004): The Grid: Blueprint for a New Computing Infrastructure (2nd Edition). Morgan-Kaufmann, San Francisco.

7. I. Foster, C. Kesselman, S. Tuecke (2001): The anatomy of the Grid: enabling scalable virtual organizations. Intl. J. Supercomputer Applications.

8. R. Hall, A.L. Rosenberg, A. Venkataramani (2006): A comparison of dag-scheduling strategies for Internet-based computing (in PDF). Typescript, Univ. Massachusetts (draft).

9. E. Korpela, D. Werthimer, D. Anderson, J. Cobb, M. Lebofsky (2000): SETI@home: massively distributed computing for SETI. In Computing in Science and Engineering (P.F. Dubois, Ed.) IEEE Computer Soc. Press, Los Alamitos, CA.

10. G. Malewicz, I. Foster, A.L. Rosenberg, M. Wilde (2006): A tool for prioritizing DAGMan jobs and its evaluation (in PDF). 15th IEEE Intl. Symp. on High-Performance Distributed Computing, 156-167.

11. G. Malewicz and A.L. Rosenberg (2005): Batch-scheduling dags for Internet-based computing (in PDF). EURO-PAR 2005, Lisbon, Portugal. In Lecture Notes in Computer Science 3648, Springer-Verlag, Berlin, 262-271.

12. G. Malewicz, A.L. Rosenberg, M. Yurkewych (2006): Toward a theory for scheduling dags in Internet-based computing (in PDF). IEEE Trans. Comput. 55, 757-768.

13. A.L. Rosenberg (2004): On scheduling mesh-structured computations for Internet-based computing (in PDF). IEEE Trans. Comput. 53, 1176-1186.

14. A.L. Rosenberg (2006): Changing challenges for collaborative algorithmics (in PDF). In Handbook of Nature-Inspired and Innovative Computing: Integrating Classical Models with Emerging Technologies (A. Zomaya, ed.) Springer-Verlag, New York, pp. 1-44.

15. A.L. Rosenberg and M. Yurkewych (2005): Guidelines for scheduling some common computation-dags for Internet-based computing (in PDF). IEEE Trans. Comput. 54, 428-438.