Multi Objective Prioritized Workflow Scheduling Using Deep Reinforcement Based Learning in Cloud Computing
Workflow Scheduling is a huge challenge in cloud paradigm as many number of workflows dynamically generated from various heterogeneous resources and task dependencies in each workflow varies from each other. Therefore, if a workflow with more number of dependencies is not scheduled onto an appropria...
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2024-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10382514/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832540571573420032 |
|---|---|
| author | Sudheer Mangalampalli Syed Shakeel Hashmi Amit Gupta Ganesh Reddy Karri K. Varada Rajkumar Tulika Chakrabarti Prasun Chakrabarti Martin Margala |
| author_facet | Sudheer Mangalampalli Syed Shakeel Hashmi Amit Gupta Ganesh Reddy Karri K. Varada Rajkumar Tulika Chakrabarti Prasun Chakrabarti Martin Margala |
| author_sort | Sudheer Mangalampalli |
| collection | DOAJ |
| description | Workflow Scheduling is a huge challenge in cloud paradigm as many number of workflows dynamically generated from various heterogeneous resources and task dependencies in each workflow varies from each other. Therefore, if a workflow with more number of dependencies is not scheduled onto an appropriate Virtual Machine i.e. with low processing capacity which leads to delay in executing workflows and it results in increase of makespan, cost, energy consumption. In order to effectively schedule complex workflows i.e. with more task dependencies, we propose a novel multi objective workflow scheduling algorithm using Deep reinforcement Learning. Initially, priorities of all workflows calculated based on their dependencies and then calculated priorities of VMs based on electricity cost at datacenters to map workflows onto precise VMs. These priorities are fed to scheduler which uses Deep Q-Network model to dynamically schedule tasks by considering both priorities of tasks and VMs. Extensive simulations carried out on workflowsim by considering realtime scientific workflows (Montage, cybershake, Epigenomics, LIGO). Our proposed MOPWSDRL compared against existing state of art approaches i.e. Heterogeneous Earliest First Deadline, Cat Swarm Optimization, Ant Colony Optimization. Results revealed that our proposed MOPDSWRL outperforms existing state of art algorithms by minimizing makespan, energy consumption. |
| format | Article |
| id | doaj-art-f71e5652577d43c1aec095e568be41a8 |
| institution | Kabale University |
| issn | 2169-3536 |
| language | English |
| publishDate | 2024-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj-art-f71e5652577d43c1aec095e568be41a82025-02-05T00:00:38ZengIEEEIEEE Access2169-35362024-01-01125373539210.1109/ACCESS.2024.335074110382514Multi Objective Prioritized Workflow Scheduling Using Deep Reinforcement Based Learning in Cloud ComputingSudheer Mangalampalli0https://orcid.org/0000-0002-1485-8783Syed Shakeel Hashmi1https://orcid.org/0000-0002-8004-4753Amit Gupta2Ganesh Reddy Karri3https://orcid.org/0000-0002-5177-8125K. Varada Rajkumar4https://orcid.org/0000-0001-5046-1276Tulika Chakrabarti5Prasun Chakrabarti6https://orcid.org/0000-0001-8062-4144Martin Margala7https://orcid.org/0000-0002-0034-0369School of Computer Science and Engineering, VIT-AP University, Amaravati, Andhra Pradesh, IndiaDepartment of Electronics and Communication Engineering, Faculty of Science and Technology (IcfaiTech), ICFAI Foundation for Higher Education (Deemed to be University), Hyderabad, Telangana, IndiaDepartment of AI & ML, J B Institute of Engineering and Technology, Hyderabad, Telangana, IndiaSchool of Computer Science and Engineering, VIT-AP University, Amaravati, Andhra Pradesh, IndiaDepartment of Computer Science and Engineering, MLR Institute of Technology, Hyderabad, Telangana, IndiaDepartment of Basic Sciences, Sir Padampat Singhania University, Udaipur, Rajasthan, IndiaDepartment of Basic Sciences, Sir Padampat Singhania University, Udaipur, Rajasthan, IndiaSchool of Computing and Informatics, University of Louisiana at Lafayette, Lafayette, LA, USAWorkflow Scheduling is a huge challenge in cloud paradigm as many number of workflows dynamically generated from various heterogeneous resources and task dependencies in each workflow varies from each other. Therefore, if a workflow with more number of dependencies is not scheduled onto an appropriate Virtual Machine i.e. with low processing capacity which leads to delay in executing workflows and it results in increase of makespan, cost, energy consumption. In order to effectively schedule complex workflows i.e. with more task dependencies, we propose a novel multi objective workflow scheduling algorithm using Deep reinforcement Learning. Initially, priorities of all workflows calculated based on their dependencies and then calculated priorities of VMs based on electricity cost at datacenters to map workflows onto precise VMs. These priorities are fed to scheduler which uses Deep Q-Network model to dynamically schedule tasks by considering both priorities of tasks and VMs. Extensive simulations carried out on workflowsim by considering realtime scientific workflows (Montage, cybershake, Epigenomics, LIGO). Our proposed MOPWSDRL compared against existing state of art approaches i.e. Heterogeneous Earliest First Deadline, Cat Swarm Optimization, Ant Colony Optimization. Results revealed that our proposed MOPDSWRL outperforms existing state of art algorithms by minimizing makespan, energy consumption.https://ieeexplore.ieee.org/document/10382514/Deep reinforcement learningcloud computingworkflow schedulingtask dependenciesmakespanenergy consumption |
| spellingShingle | Sudheer Mangalampalli Syed Shakeel Hashmi Amit Gupta Ganesh Reddy Karri K. Varada Rajkumar Tulika Chakrabarti Prasun Chakrabarti Martin Margala Multi Objective Prioritized Workflow Scheduling Using Deep Reinforcement Based Learning in Cloud Computing IEEE Access Deep reinforcement learning cloud computing workflow scheduling task dependencies makespan energy consumption |
| title | Multi Objective Prioritized Workflow Scheduling Using Deep Reinforcement Based Learning in Cloud Computing |
| title_full | Multi Objective Prioritized Workflow Scheduling Using Deep Reinforcement Based Learning in Cloud Computing |
| title_fullStr | Multi Objective Prioritized Workflow Scheduling Using Deep Reinforcement Based Learning in Cloud Computing |
| title_full_unstemmed | Multi Objective Prioritized Workflow Scheduling Using Deep Reinforcement Based Learning in Cloud Computing |
| title_short | Multi Objective Prioritized Workflow Scheduling Using Deep Reinforcement Based Learning in Cloud Computing |
| title_sort | multi objective prioritized workflow scheduling using deep reinforcement based learning in cloud computing |
| topic | Deep reinforcement learning cloud computing workflow scheduling task dependencies makespan energy consumption |
| url | https://ieeexplore.ieee.org/document/10382514/ |
| work_keys_str_mv | AT sudheermangalampalli multiobjectiveprioritizedworkflowschedulingusingdeepreinforcementbasedlearningincloudcomputing AT syedshakeelhashmi multiobjectiveprioritizedworkflowschedulingusingdeepreinforcementbasedlearningincloudcomputing AT amitgupta multiobjectiveprioritizedworkflowschedulingusingdeepreinforcementbasedlearningincloudcomputing AT ganeshreddykarri multiobjectiveprioritizedworkflowschedulingusingdeepreinforcementbasedlearningincloudcomputing AT kvaradarajkumar multiobjectiveprioritizedworkflowschedulingusingdeepreinforcementbasedlearningincloudcomputing AT tulikachakrabarti multiobjectiveprioritizedworkflowschedulingusingdeepreinforcementbasedlearningincloudcomputing AT prasunchakrabarti multiobjectiveprioritizedworkflowschedulingusingdeepreinforcementbasedlearningincloudcomputing AT martinmargala multiobjectiveprioritizedworkflowschedulingusingdeepreinforcementbasedlearningincloudcomputing |