Stochastic-based robust dynamic resource allocation for independent tasks in a heterogeneous computing system

Abstract

Heterogeneous parallel and distributed computing systems frequently must operate in environments where there is uncertainty in system parameters. Robustness can be defined as the degree to which a system can function correctly in the presence of parameter values different from those assumed. In such an environment, the execution time of any given task may fluctuate substantially due to factors such as the content of data to be processed. Determining a resource allocation that is robust against this uncertainty is an important area of research. In this study, we define a stochastic robustness measure to facilitate resource allocation decisions in a dynamic environment where tasks are subject to individual hard deadlines and each task requires some input data to start execution. In this environment, the tasks that cannot meet their deadlines are dropped (i.e., discarded). We define methods to determine the stochastic completion times of tasks in the presence of the task dropping. The stochastic task completion time is used in the definition of the stochastic robustness measure. Based on this stochastic robustness measure, we design novel resource allocation techniques that work in immediate and batch modes, with the goal of maximizing the number of tasks that meet their individual deadlines. We compare the performance of our technique against several well-known approaches taken from the literature and adapted to our environment. Simulation results of this study demonstrate the suitability of our new technique in a dynamic heterogeneous computing system.

Efficiency in the serverless cloud paradigm: A survey on the reusing and approximation aspects

“…Then, the concept was further developed to a more deliberate dynamic precision scaling 140 where calculating the precision is scaled based on multiple factors, including the trade-off between computing precision and energy or turnaround time requirement. Stochastic computing 141 is a popular collection of techniques to achieve precision scaling via representing a continuous value in form of a stream of bits. In this case, calculating the precision can be scaled by altering the number of bits in the bit stream.…”

“…166 Provided that the user functions can utilize heterogeneous resources through virtualization or heterogeneous-supported frameworks (e.g., TensorFlow for machine learning tasks, FFmpeg 168,169 for multimedia processing tasks), the challenge is how to schedule and provision such functions on the heterogeneous resources efficiently. While various forms of heterogeneous-aware task scheduler already exist 94,141,151 in the HPC context, serverless tasks are often of finer granularity. Therefore, a lightweight and low-latency scheduling that is aware of the machine heterogeneity is desired.…”

Dynamic Task Scheduling Algorithm with Deadline Constraint in Heterogeneous Volunteer Computing Platforms

“…The MOC algorithm [13] was proposed to solve the dynamic scheduling problem with deadline constraint in heterogeneous distributed computing, as mentioned before. However, the objective of MOC is that each task must be completed before its deadline, which is different from our work: To complete as many tasks as possible within their respective deadline constraints.…”

“…In addition, since there are not enough resources to complete all tasks in volunteer computing platforms, we mainly focus on the algorithm of completing as many tasks as possible before the deadline for each task. Similarly, Salehi et al proposed a maximum on-time completions (MOC) [13] algorithm for task scheduling with deadline constraint for heterogeneous distributed platforms. In the MOC algorithm, a stochastic robustness measure is defined to assign tasks, and the algorithm discards tasks that miss their deadlines to maximize the number of the completed tasks.…”

Task Assignment Algorithm Based on Trust in Volunteer Computing Platforms

“…They adopted stochastic local search to reduce the computational complexity. Furthermore, a dynamic resource assignment MOC algorithm [24] for an independent task was proposed by Salehi. The objective of the algorithm is to maximize the number of tasks completed while ensuring reliability.…”