Improving Robustness of Heterogeneous Serverless Computing Systems via Probabilistic Task Pruning

Abstract

Cloud-based serverless computing is an increasingly popular computing paradigm. In this paradigm, different services have diverse computing requirements that justify deploying an inconsistently Heterogeneous Computing (HC) system to efficiently process them. In an inconsistently HC system, each task needed for a given service, potentially exhibits different execution times on each type of machine. An ideal resource allocation system must be aware of such uncertainties in execution times and be robust against them, so that Quality of Service (QoS) requirements of users are met. This research aims to maximize the robustness of an HC system utilized to offer a serverless computing system, particularly when the system is oversubscribed. Our strategy to maximize robustness is to develop a task pruning mechanism that can be added to existing task-mapping heuristics without altering them. Pruning tasks with a low probability of meeting their deadlines improves the likelihood of other tasks meeting their deadlines, thereby increasing system robustness and overall QoS. To evaluate the impact of the pruning mechanism, we examine it on various configurations of heterogeneous and homogeneous computing systems. Evaluation results indicate a considerable improvement (up to 35%) in the system robustness.

Work-in-Progress: Evaluating Task Dropping Strategies for Overloaded Real-Time Systems

“…Job pruning in heterogeneous computing systems: In a series of publications [5], [6], [10], [17], [20], job pruning techniques have been investigated. The authors consider an oversubscribed system to which jobs are submitted at random times.…”

Efficiency in the Serverless Cloud Computing Paradigm: A Survey Study

“…In some use cases, the workflow includes optional steps (tasks) whose loss can be tolerated [121]. Such feature can be exploited at the scheduling level, particularly to mitigate resource oversubscription [35], via dropping the optional tasks [122] or deferring [35], [123] their execution to a later time when the system is less busy. One use case that can take advantage of such workflow level approximation is in video conferencing where the voice quality enhancement step (task) on the received video segments can be skipped (dropped) to keep up with the liveness of the streamed video contents [8].…”

Autonomous Task Dropping Mechanism to Achieve Robustness in Heterogeneous Computing Systems

“…Prior probabilistic task dropping approaches (e.g., [2], [16], [17]) base their dropping decisions on the chance of completing a task before its deadline (termed chance of success) and comparing that against a user-defined threshold. Nonetheless, dropping threshold is a dynamic parameter depending on system level factors, such as task arrival intensity [2].…”

“…Later, Gentry et al [2] extend the earlier study and presented a task pruning mechanism for HC systems. Denninnart et al [17], show a generalized form of the pruning mechanism and deployed it as a separate component in the system to improve robustness of homogeneous or heterogeneous systems. The generalized pruning mechanism can work in conjunction with any mapping heuristic to improve the system robustness.…”