Cost-Efficient Workflow Scheduling Algorithm for Applications With Deadline Constraint on Heterogeneous Clouds

被引:50
作者
Tang, Xiaoyong [1 ]
Cao, Wenbiao [1 ]
Tang, Huiya [2 ]
Deng, Tan [1 ]
Mei, Jing [3 ]
Liu, Yi [1 ]
Shi, Cheng [1 ]
Xia, Meng [1 ]
Zeng, Zeng [4 ]
机构
[1] Changsha Univ Sci & Technol, Sch Comp & Commun Engn, Changsha 410114, Hunan, Peoples R China
[2] Kong Baptist Univ United Int Coll UIC, Beijing Normal Univ, Appl Econ, Xiangzhou 519000, Zhuhai, Peoples R China
[3] Hunan Normal Univ, Coll Informat Sci & Engn, Changsha 410081, Hunan, Peoples R China
[4] ASTAR, I2R, Singapore 138632, Singapore
基金
中国国家自然科学基金;
关键词
Cloud computing; Task analysis; Costs; Computational modeling; Scheduling; Job shop scheduling; Heuristic algorithms; Workflow application; cost; heterogeneous clouds; schedule length; task scheduling; SCIENTIFIC WORKFLOWS; TASKS;
D O I
10.1109/TPDS.2021.3134247
中图分类号
TP301 [理论、方法];
学科分类号
081202 ;
摘要
In recent years, more and more large-scale data processing and computing workflow applications run on heterogeneous clouds. Such cloud applications with precedence-constrained tasks are usually deadline-constrained and their scheduling is an essential problem faced by cloud providers. Moreover, minimizing the workflow execution cost based on cloud billing periods is also a complex and challenging problem for clouds. In realizing this, we first model the workflow applications as I/O Data-aware Directed Acyclic Graph (DDAG), according to clouds with global storage systems. Then, we mathematically state this deadline-constrained workflow scheduling problem with the goal of minimum execution financial cost. We also prove that the time complexity of this problem is NP-hard by deducing from a multidimensional multiple-choice knapsack problem. Third, we propose a heuristic cost-efficient task scheduling strategy called CETSS, which includes workflow DDAG model building, task subdeadline initialization, greedy workflow scheduling algorithm, and task adjusting method. The greedy workflow scheduling algorithm mainly consists of dynamical task renting billing period sharing method and unscheduled task subdeadline relax technique. We perform rigorous simulations on some synthetic randomly generated applications and real-world applications, such as Epigenomics, CyberShake, and LIGO. The experimental results clearly demonstrate that our proposed heuristic CETSS outperforms the existing algorithms and can effective save the total workflow execution cost. In particular, CETSS is very suitable for large workflow applications.
引用
收藏
页码:2079 / 2092
页数:14
相关论文
共 45 条
[1]   Deadline-constrained workflow scheduling algorithms for Infrastructure as a Service Clouds [J].
Abrishami, Saeid ;
Naghibzadeh, Mahmoud ;
Epema, Dick H. J. .
FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE, 2013, 29 (01) :158-169
[2]   Cost-Driven Scheduling of Grid Workflows Using Partial Critical Paths [J].
Abrishami, Saeid ;
Naghibzadeh, Mahmoud ;
Epema, Dick H. J. .
IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS, 2012, 23 (08) :1400-1414
[3]  
[Anonymous], 2015, Amazon simple storage service documentation
[4]   Low-time complexity budget-deadline constrained workflow scheduling on heterogeneous resources [J].
Arabnejad, Hamid ;
Barbosa, Jorge G. ;
Prodan, Radu .
FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE, 2016, 55 :29-40
[5]   Scheduling deadline constrained scientific workflows on dynamically provisioned cloud resources [J].
Arabnejad, Vahid ;
Bubendorfer, Kris ;
Ng, Bryan .
FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE, 2017, 75 :348-364
[6]   Heuristics for Provisioning Services to Workflows in XaaS Clouds [J].
Cai, Zhicheng ;
Li, Xiaoping ;
Gupta, Jatinder N. D. .
IEEE TRANSACTIONS ON SERVICES COMPUTING, 2016, 9 (02) :250-263
[7]   CloudSim: a toolkit for modeling and simulation of cloud computing environments and evaluation of resource provisioning algorithms [J].
Calheiros, Rodrigo N. ;
Ranjan, Rajiv ;
Beloglazov, Anton ;
De Rose, Cesar A. F. ;
Buyya, Rajkumar .
SOFTWARE-PRACTICE & EXPERIENCE, 2011, 41 (01) :23-50
[8]   Cost Model and Analysis of Iterative MapReduce Applications for Hybrid Cloud Bursting [J].
Clemente-Castello, Francisco J. ;
Mayo, Rafael ;
Carlos Fernandez, Juan .
2017 17TH IEEE/ACM INTERNATIONAL SYMPOSIUM ON CLUSTER, CLOUD AND GRID COMPUTING (CCGRID), 2017, :858-864
[9]   Scientific workflows for computational reproducibility in the life sciences: Status, challenges and opportunities [J].
Cohen-Boulakia, Sarah ;
Belhajjame, Khalid ;
Collin, Olivier ;
Chopard, Jerome ;
Froidevaux, Christine ;
Gaignard, Alban ;
Hinsen, Konrad ;
Larmande, Pierre ;
Le Brass, Yvan ;
Lemoine, Frederic ;
Mareuil, Fabien ;
Menager, Herve ;
Pradal, Christophe ;
Blanchet, Christophe .
FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE, 2017, 75 :284-298
[10]  
Cormen T.H., 2005, INTRO ALGORITHMS, V2nd