A dynamic algorithm for cost- and deadline-constrained for scheduling workflows in cloud computing

Today cloud computing is used to address many problems because of its special features. Large scale applications usually are represented as ensemble of workflows and presented in forms of directed acyclic graphs (DAG). Efficient management of scheduling and resource provisioning is very important in IaaS clouds which consider important variables such as budget or deadline. Offering a model of on-demand resource provisioning and pay as you go are considered to be important features of such clouds. to get such purposes done, we ve introduced a dynamic algorithm which considers requested time for executing tasks in workflows, budget and deadline constraints and schedules them in one or many datacenters using selecting the best datacenter through comparing peer to peer bandwidth, distance and VM cost per hour. We ve also evaluated the proposed algorithm using CloudSim Simulator and real datasets and found out decision whether to accept or rejecting a workflow and also executing some workflows on remote datacenters may have good performance effect. Simulation results show that using a function to make these decisions can improve the algorithm performance.

Keywords:

cloud computing , scheduling , workflow