Malleable applications for scalable high performance computing
Iterative applications are known to run as slow as their slowest computational component. This paper introduces malleability, a new dynamic reconfiguration strategy to overcome this limitation. Malleability is the ability to dynamically change the data size and number of computational entities in an application. Malleability can be used by middleware to autonomously reconfigure an application in response to dynamic changes in resource availability in an architecture-aware manner, allowing applications to optimize the use of multiple processors and diverse memory hierarchies in heterogeneous environments.
The modular Internet Operating System (IOS) was extended to reconfigure applications autonomously using malleability. Two different iterative applications were made malleable. The first is used in astronomical modeling, and representative of maximum-likelihood applications was made malleable in the SALSA programming language. The second models the diffusion of heat over a two dimensional object, and is representative of applications such as partial differential equations and some types of distributed simulations. Versions of the heat application were made malleable both in SALSA and MPI. Algorithms for concurrent data redistribution are given for each type of application. Results show that using malleability for reconfiguration is 10 to 100 times faster on the tested environments. The algorithms are also shown to be highly scalable with respect to the quantity of data involved. While previous work has shown the utility of dynamically reconfigurable applications using only computational component migration, malleability is shown to provide up to a 15% speedup over component migration alone on a dynamic cluster environment.
This work is part of an ongoing research effort to enable applications to be highly reconfigurable and autonomously modifiable by middleware in order to efficiently utilize distributed environments. Grid computing environments are becoming increasingly heterogeneous and dynamic, placing new demands on applications’ adaptive behavior. This work shows that malleability is a key aspect in enabling effective dynamic reconfiguration of iterative applications in these environments.
KeywordsHigh performance computing Malleability Dynamic reconfiguration MPI SALSA Actors
Unable to display preview. Download preview PDF.
- 1.Agbaria, A., Friedman, R.: Starfish: Fault-tolerant dynamic MPI programs on clusters of workstations. In: Proceedings of The Eighth IEEE International Symposium on High Performance Distributed Computing, p. 31. IEEE Computer Society (1999) Google Scholar
- 2.Agha, G.: Actors: A Model of Concurrent Computation in Distributed Systems. MIT Press (1986) Google Scholar
- 4.Berman, F., Chien, A., Cooper, K., Dongarra, J., Foster, I., Gannon, D., Johnson, L., Kennedy, K., Kesselman, C., Mellor-Crummey, J., Reed, D., Torczon, L., Wolski, R.: The GrADS project: Software support for high-level grid application development. Int. J. High-Perform. Comput. Appl. 15(4), 327–344 (2002) CrossRefGoogle Scholar
- 5.Bhandarkar, M.A., Kale, L.V., de Sturler, E., Hoeflinger, J.: Adaptive load balancing for MPI programs. In: Proceedings of the International Conference on Computational Science—Part II, pp. 108–117. Springer (2001) Google Scholar
- 6.Blumofe, R.D., Leiserson, C.E.: Scheduling multithreaded computations by work stealing. In: Proceedings of the 35th Annual Symposium on Foundations of Computer Science (FOCS ’94), Santa Fe, New Mexico, November 1994, pp. 356–368 Google Scholar
- 7.Czajkowski, K., Fitzgerald, S., Foster, I., Kesselman, C.: Grid information services for distributed resource sharing. In: 10th IEEE International Symposium on High-Performance Distributed Computing (HPDC-10), August 2001 Google Scholar
- 8.Desell, T., Maghraoui, K.E., Varela, C.: Load balancing of autonomous actors over dynamic networks. In: Proceedings of the Hawaii International Conference on System Sciences, HICSS-37 Software Technology Track, January 2004, pp. 1–10 Google Scholar
- 9.Foster, I., Kesselman, C.: The Globus project: A status report. In: Antonio, J. (ed.) Proceedings of the Seventh Heterogeneous Computing Workshop (HCW ’98), pp. 4–18. IEEE Computer Society (1998) Google Scholar
- 10.Huang, C., Lawlor, O., Kalé, L.V.: Adaptive MPI. In: Proceedings of the 16th International Workshop on Languages and Compilers for Parallel Computing (LCPC 03), College Station, Texas, October 2003 Google Scholar
- 11.Lan, Z., Taylor, V.E., Bryan, G.: Dynamic load balancing of SAMR applications on distributed systems. Sci. Progr. 10(4), 319–328 (2002) Google Scholar
- 12.Litzkow, M., Livny, M., Mutka, M.: Condor—a hunter of idle workstations. In: Proceedings of the 8th International Conference of Distributed Computing Systems, June 1988, pp. 104–111 Google Scholar
- 13.Maghraoui, K.E., Flaherty, J., Szymanski, B., Teresco, J., Varela, C.: Adaptive computation over dynamic and heterogeneous networks. In: Wyrzykowski, R., Dongarra, J., Paprzycki, M., Wasniewski, J. (eds.) Proc. of the Fifth International Conference on Parallel Processing and Applied Mathematics (PPAM’2003), Czestochowa, Poland, September 2003. Lecture Notes in Computer Science, vol. 3019, pp. 1083–1090. Springer, Berlin (2003) Google Scholar
- 14.Maghraoui, K.E., Desell, T., Szymanski, B.K., Teresco, J.D., Varela, C.A.: Towards a middleware framework for dynamically reconfigurable scientific computing. In: Grandinetti, L. (ed.) Grid Computing and New Frontiers of High Performance Processing. Elsevier (2005) Google Scholar
- 15.Maghraoui, K.E., Desell, T.J., Szymanski, B.K., Varela, C.A.: The internet operating system: Middleware for adaptive distributed computing. Int. J. High Perform. Comput. Appl. (IJHPCA) 10(4), 467–480 (2006), Special Issue on Scheduling Techniques for Large-Scale Distributed Platforms CrossRefGoogle Scholar
- 16.Maghraoui, K.E., Szymanski, B., Varela, C.: An architecture for reconfigurable iterative mpi applications in dynamic environments. In: Wyrzykowski, R., Dongarra, J., Meyer, N., Wasniewski, J. (eds.) Proc. of the Sixth International Conference on Parallel Processing and Applied Mathematics (PPAM’2005), ser. LNCS, no. 3911, Poznan, Poland, September 2005, pp. 258–271 Google Scholar
- 17.Message Passing Interface Forum, MPI: A message-passing interface standard, Int. J. Supercomput. Appl. High Perform. Comput. 8(3/4), 159–416 (1994) Google Scholar
- 19.Purnell, J., Magdon-Ismail, M., Newberg, H.: A probabilistic approach to finding geometric objects in spatial datasets of the Milky Way. In: Proceedings of the 15th International Symposium on Methodoligies for Intelligent Systems (ISMIS 2005), Saratoga Springs, NY, USA, May 2005, pp. 475–484. Springer (2005) Google Scholar
- 21.Stellner, G.: Cocheck: Checkpointing and process migration for MPI. In: Proceedings of the 10th International Parallel Processing Symposium, pp. 526–531. IEEE Computer Society (1996) Google Scholar
- 23.Taura, K., Kaneda, K., Endo, T.: Phoenix: a parallel programming model for accommodating dynamically joininig/leaving resources. In: Proc. of PPoPP, pp. 216–229. ACM (2003) Google Scholar
- 25.Varela, C., Agha, G.: Programming dynamically reconfigurable open systems with SALSA. ACM SIGPLAN Not. OOPSLA’2001 Intriguing Techn. Track Proc. 36(12), 20–34 (2001), http://www.cs.rpi.edu/~cvarela/oopsla2001.pdf Google Scholar
- 26.Varela, C.A., Ciancarini, P., Taura, K.: Worldwide computing: Adaptive middleware and programming technology for dynamic Grid environments. Sci. Program. J. 13(4), 255–263 (2005), Guest Editorial Google Scholar