Online scheduling of dependent tasks of cloud’s workflows to enhance resource utilization and reduce the makespan using multiple reinforcement learning-based agents
“…In the second group, we found studies presenting approaches for workflow schedules that calculate makespan using an estimation of execution time, considering both stochastic and constant variables 13,38‐41 . Other approaches use heuristics and simulations for measuring, analysing and reducing makespan as well as improving the running processes 42‐51 …”
Section: Related Workmentioning
confidence: 99%
“…Similarly, Aziz et al 49 proposed an algorithm that deals with unsuccessful task execution by dynamically scheduling workflows to available resources aiming to optimise the makespan and minimise the reliability redistributing the failed task to nonused resources. The study presented by Asghari et al 50 proposes an algorithm based on multi‐agent system for task scheduling and resource provisioning focused on reducing makespan, minimise power, optimise the cost of using the resources, and maximise the utilisation of the resources. The authors used a learning agent‐based resource management framework for resource provisioning to the tasks.…”
Increasingly enterprises rely on software applications to support their business processes. Since such processes are continually evolving to keep up with market dynamism, companies strive to increase their efficiency, for example, by optimising the integration of applications supporting these processes. Integration platforms are specialised software tools that allow creating integration processes so that applications can share data and functionality. However, this integration involves several challenges, especially when large volumes of heterogeneous data should be integrated and shared. The performance of an integration process, in terms of message processing, is directly related to the run-time system of the integration platform. This article investigates the impact of the volume of messages and the number of threads used by a run-time system on makespan and performance of an integration process. The greater is the number of messages per second received by the integration process, the high is the volume of messages. The study was based on a run-time system with task-based execution model and follows a strict protocol to conduct and report our empirical study. We observed an increment of makespan when increasing the volume of messages to integration processes and different behaviours when increasing the number of threads used in their executions. Makespan reduces as the number of threads increases, but only when the volume of inbound messages is not very high. We confirmed that there is a performance gain by increasing the number of threads to execute an integration process, but observed that the continuous increment of threads leads to degradation of the performance in this model.
“…In the second group, we found studies presenting approaches for workflow schedules that calculate makespan using an estimation of execution time, considering both stochastic and constant variables 13,38‐41 . Other approaches use heuristics and simulations for measuring, analysing and reducing makespan as well as improving the running processes 42‐51 …”
Section: Related Workmentioning
confidence: 99%
“…Similarly, Aziz et al 49 proposed an algorithm that deals with unsuccessful task execution by dynamically scheduling workflows to available resources aiming to optimise the makespan and minimise the reliability redistributing the failed task to nonused resources. The study presented by Asghari et al 50 proposes an algorithm based on multi‐agent system for task scheduling and resource provisioning focused on reducing makespan, minimise power, optimise the cost of using the resources, and maximise the utilisation of the resources. The authors used a learning agent‐based resource management framework for resource provisioning to the tasks.…”
Increasingly enterprises rely on software applications to support their business processes. Since such processes are continually evolving to keep up with market dynamism, companies strive to increase their efficiency, for example, by optimising the integration of applications supporting these processes. Integration platforms are specialised software tools that allow creating integration processes so that applications can share data and functionality. However, this integration involves several challenges, especially when large volumes of heterogeneous data should be integrated and shared. The performance of an integration process, in terms of message processing, is directly related to the run-time system of the integration platform. This article investigates the impact of the volume of messages and the number of threads used by a run-time system on makespan and performance of an integration process. The greater is the number of messages per second received by the integration process, the high is the volume of messages. The study was based on a run-time system with task-based execution model and follows a strict protocol to conduct and report our empirical study. We observed an increment of makespan when increasing the volume of messages to integration processes and different behaviours when increasing the number of threads used in their executions. Makespan reduces as the number of threads increases, but only when the volume of inbound messages is not very high. We confirmed that there is a performance gain by increasing the number of threads to execute an integration process, but observed that the continuous increment of threads leads to degradation of the performance in this model.
“…Machine learning-based approaches have hence attracted lots of attention in the recent decade [15] [16] [35]. Machine learning approaches are tried to handle makespan of task flows [36], resource utilization rate [37], Quality of Service [38] and pricing models [39].…”
“…e algorithm was not limited to adapt to its own task arrival process but also fully considered the influence of other agents on the task flow. Asghari et al [33] proposed a RL-based resource allocation method in order to reduce the cost of system and improve the utilization of resource. Wauters et al [34] developed a learning-based resource scheduling optimization method to minimize the average delay and total completion time of the project.…”
As one of the most effective medical technologies for the infertile patients, in vitro fertilization (IVF) has been more and more widely developed in recent years. However, prolonged waiting for IVF procedures has become a problem of great concern, since this technology is only mastered by the large general hospitals. To deal with the insufficiency of IVF service capacity, this paper studies an IVF queuing network in an integrated cloud healthcare system, where the two key medical services, that is, egg retrieval and transplantation, are assigned to accomplish in the general hospital, while the routine medical tests are assigned into the community hospital. Based on continuous-time Markov procedure, a dynamic large-scale server scheduling problem in this complicated service network is modeled with consideration of different arrival rates of multiple type of patients and different service capacities of multiple servers that can be defined as doctors of the general hospital. To solve this model, a reinforcement learning (RL) algorithm is proposed, where the reward functions are designed for four conflicting subcosts: setup cost, patient waiting cost, penalty cost for unsatisfied patient personal preferences, and medical cost of patient. The experimental results show that the optimal service rule of each server’s queue obtained by the RL method is significantly superior to the traditional service rule.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.