Scheduling on Large Scale Distributed Platforms: From Models to Implementations

Abstract

Today, large scale parallel systems are available at low cost. Many powerful such systems have been installed all over the world and the number of users is always increasing. The difficulty of using them efficiently is growing with the complexity of the interactions between more and more architectural constraints and the diversity of the applications. The design of efficient parallel algorithms has to be reconsidered under the influence of new parameters of such platforms (namely, cluster, grid and global computing) which are characterized by a larger number of heterogeneous processors, often organized in several hierarchical sub-systems. At each step of the evolution of the parallel processing field, researchers designed adequate computational models whose objective was to abstract the real world in order to be able to analyze the behavior of algorithms.In this paper, we will investigate two complementary computational models that have been proposed recently: Parallel Task (PT) and Divisible Load (DL). The Parallel Task (i.e. tasks that require more than one processor for their execution) model is a promising alternative for scheduling parallel applications, especially in the case of slow communication media. The basic idea is to consider the application at a coarse level of granularity. Another way of looking at the problem (which is somehow a dual view) is the Divisible Load model where an application is considered as a collection of a large number of elementary -sequential -computing units that will be distributed among the available resources. Unlike the PT model, the DL model corresponds to a fine level of granularity. We will focus on the PT model, and discuss how to mix it with simple Divisible Load scheduling.As the main difficulty for distributing the load among the processors (usually known as the scheduling problem) in actual systems comes from handling efficiently the communications, these two models of the problem allow us to consider them implicitly or to mask them, thus leading to more tractable problems.We will show that in spite of the enormous complexity of the general scheduling problem on new platforms, it is still useful to study theoretical 1 Contact Author: Denis Trystram (Denis.Trystram@imag.fr) 1 models. We will focus on the links between models and actual implementations on a regional grid with more than 500 processors.

A multi-criteria job scheduling framework for large computing farms

“…Even if the multi-criteria approach seems to be a proper technique to efficiently solve the scheduling problem on heterogeneous and distributed computing platforms, only a few research solutions have been proposed. In [24,7,8] a bicriteria algorithm for scheduling jobs on clusters is shown. It exploits two pre-selected criteria for minimizing job makespan as well as their average completion time.…”

“…The heuristics tries to minimize job migrations by increasing the priorities of the job-machine associations present in the current scheduling plan. p i,m is computed as: (8) where W C is the heuristics weight, and T elapsed i,m is the elapsed execution time of a job i on machine m.…”

“…For each simulation, we randomly generated a list of jobs and machines whose parameters are generated according to a uniform distribution in the ranges: estimated [500-3000], deadline , number of CPUs [1][2][3][4][5][6][7][8], benchmark [200-600], license ratio [55%-65%], license suitability 90%, license needs 20%.…”

“…3) fixed equal to 1, 4, 6, 8, 10, and 12 simulator-time units (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)5,6,7,8,9).…”

Checkpointing Strategies for Scheduling Computational Workflows

“…In addition, complex data redistributions must take place so that output data from one task can serve as input data to another task when the two tasks do not use the same number of processors. It is not clear how to model redistribution costs in practice and thus how to make judicious scheduling and processor allocation decisions [11,12]. Because we consider processor failures, which makes the scheduling problem even more difficult, in this work we opt for a simplified scenario in which each task uses all the available processors.…”