Deadline‐constrained cost‐energy aware workflow scheduling in cloud

Abstract

Nowadays, scientists are dealing with large-scale scientific workflows that need a high processing capacity platform to facilitate on-time completion. Cloud computing is the ideal platform to overcome this problem as it has several resources that scientists may choose from depending on the size of their applications. However, using cloud computing requires some monetary charges. Recently cloud computing providers started a new pricing schema that offers to their users a set of resources with specific combinations of CPU frequency configurations settings and price. The selected configurations settings reflect energy consumption. Besides, the configuration selection to meet users' satisfaction (minimum cost) and providers' satisfaction (energy saving) is crucial. Therefore, a multiobjective (cost and energy) efficient mechanism is essential. In this article, we address an important novel problem concerning multiobjective deadline constrained workflow scheduling in the cloud. We first study the relationship between cost minimization and minimization of the energy consumption in a cloud environment, and then discuss, develop, and propose an algorithm with two variants to help the system satisfy both sides (users and providers) at the same time during the selection of the configuration. The proposed heuristic is evaluated using specified real-world applications. The observed results indicate that our heuristic can reduce significantly the energy consumption and the cost at the same time. K E Y W O R D Scloud computing, cost-efficient workflow scheduling, deadline constrained workflow scheduling, energy efficient workflow scheduling, multicriteria workflow scheduling, multiobjective workflow scheduling, weighted sum for workflow scheduling INTRODUCTIONLarge-scale workflow-based applications appear in most of every scientific experiment. They are presented with many interdependent tasks, computing dependencies, and data flows. Workflow scheduling is the process of mapping workflow tasks to the available computing resources for the satisfaction of some constraints. Among distributed systems, Cloud computing has been witnessed to be the applicable environment to handle the computation requirements of large-scale workflow-based applications. Cloud computing is the lasted generation of the distributed system like grid computing, cluster computing, and parallel computing, which offers to its user a resizable computing resource, storage, and bandwidth on a pay

Workflow Scheduling With Guaranteed Responsiveness and Minimal Cost

“…First, the tasks on the critical path are assigned to the least cost virtual machines while meeting the workflow deadline. The [17] Cost × N/A CPI [18] Cost × O(n 3 d 2 m) LBWS & SAWS [13] Cost × O(n 2 k) CETSS [19] Cost × O(n 3 ) CIPS & LHCM [20] Cost × O(n 3 D 2 W ) CPDE [21] Cost × O((m + m1)n 2 ) ILAH [22] Cost × O(n 2 ln n) PCP-B 2 [23] Makespan × O(n 2 log m) CB-DT [24] Makespan × O(n 2 ) AILS [25] Makespan × N/A DCCP [26] Cost × O(n 2 p) RCT/RTC [27] Makespan and Cost × × O(n 2 ) MLCP [28] Makespan × N/A BDAS [30] Makespan and Cost × × N/A DCHG-TS [31] Makespan and Cost × × Metaheuristic Calzarossa et al [29] Makespan and Cost × × Metaheuristic BDCE & BDD [32] Cost and Energy × × O(n 2 ) Bugingo et al [33] Cost and Energy × N/A ECWS [34] Cost, Energy, and × O(n 2 v) Resource Utilization CEAS [35] Cost and Energy × O(Kn 2 (n + e)) EViMA [36] Makespan, Cost, and × N/A Energy CTDC [12] Cost × O(n 2 ) MTDC [12] Makespan and Cost O(n 2 ) CEFA [37] Makespan and Cost × Metaheuristic handled path is then deleted and the resulting blocks (subworkflows) are considered. Mapping a workflow to an algebraic structure with a totally order set (T exe (P), ) helps to preserve the logical relation between the main workflow and sub-workflows in each step.…”

“…Multi-Objective and Unconstrained Optimization: Some of the published studies considered more than one metric in optimizing the workflow schedule with and without factoring in time constraints. Bugingo et al opt to minimize cost and energy in a cloud environment [33]. A similar variant of the scheduling problem is tackled in [34], by optimizing energy consumption, execution cost, and resource utilization in cloud data centers.…”

TEBAS: A Time-Efficient Balance-Aware Scheduling Strategy for Batch Processing Jobs

“…To reduce resource contention between co-located tasks, this paper attempts to find the optimal task assignment solution X * from X = [X n t ] T j s ×n which achieves the minimum maxUtil under several constraints from (5) to (9).…”

“…Several greedy strategies are proposed in [2] to minimize cost consumption within the deadline by tuning CPU frequency for BP jobs based on the initial placement. To extend this, a comparative analysis of multiple initial place- ment heuristics is conducted in [3] to figure out how initial placement affects the greedy strategies above. ESMS [4] is an elastic scheduling algorithm which makes redundant resource configuration for upcoming jobs to minimize the financial cost of cloud resources under deadline constraints.…”

“…In [7], an improved genetic algorithm called DCHG-TS with customized genetic operators is proposed to co-optimize cost and makespan. In [8] and [9], Bugingo et al leverage the multi-decision making algorithm with equal weight and specified weight, respectively, to co-optimize the energy and financial cost. However, the above scheduling strategies perform time-consuming objective optimization which yields nonnegligible scheduling delay for short-living BP jobs.…”